A distributed fiber optic temperature measurement and fire suppression control system for mining applications

Abstract

This application relates to the field of intelligent control, specifically disclosing a distributed fiber optic temperature measurement and fire suppression control system for mines. It uses a deep learning neural network model to mine the temporal distribution correlation characteristics of temperature at various locations in key parts of the mine and uses this to adaptively control the fire suppression equipment at each location. This allows for a better understanding and grasp of temperature changes at various locations in the mine, thereby providing a more reliable and accurate basis for subsequent automated fire suppression decisions.

Description

Technical Field

[0001] This application relates to the field of intelligent control, and more specifically, to a distributed fiber optic temperature measurement and fire prevention control system for mining applications. Background Technology

[0002] Mines are typically enclosed spaces containing flammable and explosive materials. Fires in mines are extremely difficult to extinguish and rescue, potentially leading to very serious consequences. Therefore, implementing fire prevention and control measures in mines is of paramount importance.

[0003] For example, Chinese patent application number 201610919731.5 discloses a mine explosion monitoring, alarm, and control system. The system mainly includes an information processing server, an alarm device, a communication network, explosion suppression and fire extinguishing equipment, a gas concentration monitoring device, and various environmental monitoring devices. The system can monitor changes in various data such as smoke and temperature caused by a gas explosion, monitor the concentration of a marker gas through the gas concentration monitoring device, issue an alarm for a mine explosion based on the monitored data, and automatically suppress and extinguish the explosion, reducing casualties and losses caused by gas explosions. This system overcomes the shortcomings of traditional explosion monitoring methods such as gas monitoring, which suffer from slow response, high false alarm rates, and high missed alarm rates, greatly improving alarm accuracy and providing important protection for safe coal mine production.

[0004] For example, Chinese patent application number 202010542965.9 discloses an IoT-based regional fire suppression and security monitoring system, including a regional fire suppression device group, a video monitoring group, and a centralized monitoring system. The regional fire suppression device group consists of eight upgraded and modified mining-grade automatic powder spraying fire suppression devices, deployed in various electromechanical equipment chambers underground in coal mines. The video monitoring group consists of 16 intrinsically safe mining cameras, deployed in various electromechanical equipment chambers underground in coal mines, collecting video images from these chambers to provide video data for fire image recognition. The centralized monitoring system includes image monitoring and equipment parameter monitoring. Image monitoring, combined with image recognition algorithms, can analyze in real time whether a fire has occurred in the images, while equipment parameter monitoring helps staff remotely understand the equipment status. This invention can monitor whether a fire has occurred underground in coal mines from multiple dimensions. Once a fire is detected, the fire suppression devices are quickly activated, achieving remote centralized monitoring of the fire suppression equipment and minimizing the occurrence of fires underground in coal mines.

[0005] Similar to the existing fire prevention and extinguishing solutions for mines, these solutions cannot achieve full coverage, which makes it impossible to monitor temperature changes in various areas of the mine and issue timely alarms.

[0006] Therefore, there is a need for an optimized fire prevention and control scheme for mining environments. Summary of the Invention

[0007] To address the aforementioned technical problems, this application is proposed. An embodiment of this application provides a distributed fiber optic temperature measurement and fire suppression control system for mines. This system utilizes a deep learning neural network model to mine the temporal distribution correlation characteristics of temperature at various locations in key mine areas and adaptively controls the fire suppression equipment at each location. This allows for a better understanding and grasp of temperature changes at various locations within the mine, thereby providing a more reliable and accurate basis for subsequent automated fire suppression decisions.

[0008] According to one aspect of this application, a distributed fiber optic temperature measurement and fire prevention control system for mining is provided, comprising:

[0009] The temperature data acquisition module is used to acquire the discrete time-series temperature distribution of multiple locations in key parts of the mine, collected by the distributed fiber optic temperature measurement system.

[0010] The structured module is used to arrange the discrete temperature time distributions at each location into multiple temperature time input vectors according to the time dimension.

[0011] The temperature time series feature extraction module is used to pass the multiple temperature time series input vectors through a temperature time series feature extractor containing a first convolutional layer and a second convolutional layer to obtain multiple temperature time series feature vectors, wherein the first convolutional layer and the second convolutional layer use one-dimensional convolutional kernels with different scales.

[0012] The temperature time-series neighborhood correlation feature extraction module is used to arrange the multiple temperature time-series feature vectors into a two-dimensional feature matrix and then use a convolutional neural network model as a feature extractor to obtain a global temperature distribution feature matrix.

[0013] The query module is used to calculate the matrix product between each temperature time-series feature vector and the global temperature distribution feature matrix to obtain multiple classification feature vectors.

[0014] The optimization module is used to perform regularization enhancement on each of the classification feature vectors to obtain multiple optimized classification feature vectors; and

[0015] The control result generation module is used to pass the optimized classification feature vectors through a classifier to obtain classification results, which are used to indicate whether to automatically activate the fire extinguishing equipment at each location.

[0016] In the above-mentioned distributed optical fiber temperature measurement and fire prevention control system for mines, the first convolutional layer and the second convolutional layer are parallel, and the temperature time series feature extractor also includes a multi-scale feature fusion layer that is simultaneously connected to the first convolutional layer and the second convolutional layer.

[0017] In the aforementioned distributed optical fiber temperature measurement and fire suppression control system for mines, the temperature time-series feature extraction module includes: a first temperature time-series feature extraction unit, used to input the temperature time-series input vector into the first convolutional layer of the temperature time-series feature extractor to obtain a first-scale temperature time-series feature vector, wherein the first convolutional layer has a first one-dimensional convolutional kernel of a first length; a second temperature time-series feature extraction unit, used to input the temperature time-series input vector into the second convolutional layer of the temperature time-series feature extractor to obtain a second-scale temperature time-series feature vector, wherein the second convolutional layer has a second one-dimensional convolutional kernel of a second length, the first length being different from the second length; and a multi-scale cascading unit, used to cascade the first-scale temperature time-series feature vector and the second-scale temperature time-series feature vector to obtain the temperature time-series feature vector. The first temperature time-series feature extraction unit is used to: perform one-dimensional convolutional encoding on the temperature time-series input vector using the first convolutional layer of the temperature time-series feature extractor with the following one-dimensional convolution formula to obtain the first-scale temperature time-series feature vector; wherein the formula is:

[0018]

[0019] Where, a is the width of the first convolutional kernel in the x-direction, F(a) is the parameter vector of the first convolutional kernel, G(xa) is the local vector matrix operated with the convolutional kernel function, w is the size of the first convolutional kernel, X represents the temperature time-series input vector, and Cov(X) represents one-dimensional convolutional encoding of the temperature time-series input vector; and, the second temperature time-series feature extraction unit is used to: use the second convolutional layer of the temperature time-series feature extractor to perform one-dimensional convolutional encoding on the temperature time-series input vector using the following one-dimensional convolution formula to obtain the second-scale temperature time-series feature vector; wherein, the formula is:

[0020]

[0021] Where b is the width of the second convolution kernel in the x direction, F(b) is the parameter vector of the second convolution kernel, G(xb) is the local vector matrix operated with the convolution kernel function, m is the size of the second convolution kernel, X represents the temperature time series input vector, and Cov(X) represents one-dimensional convolution encoding of the temperature time series input vector.

[0022] In the aforementioned distributed optical fiber temperature measurement and fire prevention control system for mines, the temperature time-series neighborhood correlation feature extraction module is used to: use each layer of the convolutional neural network model, which serves as the feature extractor, to perform the following on the input data during the forward propagation of the layer: convolution processing on the input data to obtain a convolutional feature map; pooling along the channel dimension of the convolutional feature map to obtain a pooled feature map; and nonlinear activation on the pooled feature map to obtain an activation feature map; wherein, the output of the last layer of the convolutional neural network serving as the feature extractor is the global temperature distribution feature matrix, and the input of the first layer of the convolutional neural network serving as the feature extractor is a two-dimensional feature matrix formed by arranging the multiple temperature time-series feature vectors.

[0023] In the aforementioned distributed optical fiber temperature measurement and fire prevention control system for mines, the convolutional neural network model used as the feature extractor is a deep residual network model.

[0024] In the aforementioned distributed fiber optic temperature measurement and fire suppression control system for mines, the optimization module is used to: regularize and strengthen each classification feature vector using the following optimization formula to obtain multiple optimized classification feature vectors; wherein, the formula is:

[0025] v′ i =(μσ)v i 2 +v i μ+(v i -σ)μ 2

[0026] Where σ and v are the mean and standard deviation of the feature values ​​at each position of the classification feature vector, and v i ′ is the feature value at the i-th position of the optimized classification feature vector.

[0027] In the above-mentioned distributed optical fiber temperature measurement and fire prevention control system for mines, the control result generation module includes: a fully connected encoding unit, used to perform fully connected encoding on the optimized classification feature vector using multiple fully connected layers of the classifier to obtain an encoded classification feature vector; and a classification result generation unit, used to pass the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

[0028] According to another aspect of this application, a method for fire prevention and extinguishing using distributed optical fiber temperature measurement in mines is provided, which includes: acquiring the discrete temperature time distribution of multiple locations in key parts of the mine collected by a distributed optical fiber temperature measurement system.

[0029] The discrete temperature time distributions at each location are arranged into multiple temperature time-series input vectors according to the time dimension.

[0030] The multiple temperature time-series input vectors are respectively passed through a temperature time-series feature extractor containing a first convolutional layer and a second convolutional layer to obtain multiple temperature time-series feature vectors, wherein the first convolutional layer and the second convolutional layer respectively use one-dimensional convolutional kernels with different scales;

[0031] After arranging the multiple temperature time-series feature vectors into a two-dimensional feature matrix, a convolutional neural network model, acting as a feature extractor, is used to obtain the global temperature distribution feature matrix.

[0032] Using each temperature time-series feature vector as a query feature vector, calculate the matrix product between it and the global temperature distribution feature matrix to obtain multiple classification feature vectors;

[0033] Each of the classification feature vectors is subjected to regularization enhancement to obtain multiple optimized classification feature vectors; and

[0034] The optimized classification feature vectors are passed through a classifier to obtain classification results, which are used to indicate whether the fire extinguishing equipment at each location is automatically activated.

[0035] According to another aspect of this application, an electronic device is provided, comprising: a processor; and a memory storing computer program instructions, which, when executed by the processor, cause the processor to perform the mining distributed fiber optic temperature measurement and fire suppression control method as described above.

[0036] According to another aspect of this application, a computer-readable medium is provided having computer program instructions stored thereon, which, when executed by a processor, cause the processor to perform the mining distributed fiber optic temperature measurement and fire prevention control method as described above.

[0037] Compared with existing technologies, the distributed fiber optic temperature measurement and fire suppression control system for mines provided in this application uses a deep learning neural network model to mine the temporal distribution correlation characteristics of temperature at various locations in key parts of the mine and adaptively controls the fire suppression equipment at each location. In this way, it is possible to better understand and grasp the temperature changes at various locations in the mine, thereby providing a more reliable and accurate basis for subsequent automated fire suppression decisions. Attached Figure Description

[0038] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0039] Figure 1 This is a block diagram of a mine-use distributed optical fiber temperature measurement and fire prevention control system according to an embodiment of this application;

[0040] Figure 2 This is a system architecture diagram of a mine-use distributed optical fiber temperature measurement and fire prevention control system according to an embodiment of this application.

[0041] Figure 3 A block diagram of a temperature time-series feature extraction module in a mine-use distributed optical fiber temperature measurement and fire extinguishing control system according to an embodiment of this application;

[0042] Figure 4 This is a flowchart of the convolutional neural network encoding in the mine distributed optical fiber temperature measurement and fire extinguishing control system according to an embodiment of this application;

[0043] Figure 5 This is a flowchart of a mining distributed optical fiber temperature measurement and fire prevention control method according to an embodiment of this application;

[0044] Figure 6 This is a block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0045] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0046] Application Overview

[0047] To address the aforementioned technical problems, the technical concept of this application is to utilize a distributed optical fiber temperature measurement system (DTS) to monitor the temperature in various areas of the mine, thereby promptly detecting fire hazards and automatically activating fire extinguishing equipment.

[0048] Specifically, in the technical solution of this application, the discrete temporal distribution of temperature at multiple locations in key parts of the mine is first obtained by a distributed fiber optic temperature measurement system. As mentioned earlier, mines contain a large number of flammable, explosive, and other hazardous materials. These materials will ignite and cause fires once stimulated by high temperatures or flames. Therefore, monitoring and early warning of temperature changes in various areas of the mine is an important means of preventing mine fires.

[0049] Distributed fiber optic temperature measurement (DTS) systems can collect time-series temperature data at various locations within a mine and transmit the data to a subsequent processing system for analysis and processing. This acquired temperature time-series distribution data allows for feature extraction and classification in subsequent technical solutions, thereby achieving the goal of detecting temperature changes at various locations within the mine and automatically activating fire suppression equipment. Specifically, utilizing the basic principle of the DTS system—that is, using the ratio of Raman scattered light intensity generated by the interaction between a laser pulse and fiber optic molecules—the temperature values ​​at different locations along the fiber are calculated. The location information is determined based on the propagation time of the light pulse in the fiber. This allows for the acquisition of the time-series temperature distribution at multiple locations (e.g., every 0.5 meters) in key mine components (such as cables, conveyor belts, and fans), i.e., the temperature value at each location at different times.

[0050] Next, the discrete temperature time distributions at each location are arranged into multiple temperature time-series input vectors according to the time dimension. That is, the data format of the discrete temperature time distributions at each location is adjusted to transform the original temperature data into an input format suitable for deep learning processing. In other words, the temperature time distribution at each location is regarded as a one-dimensional array, where each element represents the temperature value at a time point. This results in multiple one-dimensional arrays of the same length (e.g., 1000), i.e., multiple temperature time-series input vectors.

[0051] Furthermore, the multiple temperature-time input vectors are respectively passed through a temperature-time feature extractor containing a first convolutional layer and a second convolutional layer to obtain multiple temperature-time feature vectors, wherein the first convolutional layer and the second convolutional layer respectively use one-dimensional convolutional kernels with different scales. Here, the temperature-time feature extractor is used to perform multi-scale one-dimensional convolutional encoding on each temperature-time input vector to capture the temperature-time distribution features of multiple locations in the key parts of the mine.

[0052] It should be understood that in the technical solution of this application, the temperature change patterns at different locations during a fire in a mine may vary significantly. Therefore, feature extraction and classification of temperature time-series data for different locations are necessary. Since temperature time-series data inherently possesses time-series properties, convolutional neural networks can be used for feature extraction and classification. Using one-dimensional convolutional kernels of different scales for convolution operations can better adapt to temperature time-series data at different locations, thereby improving the accuracy of feature extraction and classification. Specifically, the first convolutional layer uses a smaller convolutional kernel to quickly capture local temperature change features and extract more refined local feature information; while the second convolutional layer uses a larger convolutional kernel to better capture more global temperature change features and extract more macroscopic global feature information. This organically combines local and global feature information to form a more complete and accurate temperature time-series feature vector, providing a more reliable foundation for subsequent classification decisions.

[0053] Next, the multiple temperature time-series feature vectors are arranged into a two-dimensional feature matrix, and then a convolutional neural network model, acting as a feature extractor, is used to obtain a global temperature distribution feature matrix. That is, the temperature time-series distribution features of each location in the key parts of the mine are integrated into a global temperature distribution matrix, and a convolutional neural network model is used to perform local neighborhood feature extraction based on convolution kernels on the global temperature distribution matrix to capture the correlation pattern features between the temperature time-series distributions of each location in the key parts of the mine.

[0054] Specifically, the temperature time-series feature vectors of each location are arranged into a two-dimensional feature matrix along the time dimension, where each row represents the temperature time-series feature vector of a location at different time points. This two-dimensional feature matrix is ​​then input into a convolutional neural network model, which acts as a feature extractor, to obtain a feature matrix containing global temperature distribution information. The convolutional neural network model can perform convolution and pooling operations on the input two-dimensional feature matrix to extract the correlations and patterns between different locations in the mine. In this way, the connections and differences between the temperature time-series feature vectors of various locations can be fully explored and integrated into a global temperature distribution feature matrix. This allows for a better understanding and grasp of temperature changes at various locations in the mine, providing a more comprehensive and accurate reference for subsequent classification decisions.

[0055] Next, using each temperature time-series feature vector as a query feature vector, the matrix product between it and the global temperature distribution feature matrix is ​​calculated to obtain multiple classification feature vectors. Here, using each temperature time-series feature vector as a query feature vector and calculating its matrix product with the global temperature distribution feature matrix to obtain multiple classification feature vectors is to compare the temperature time-series data at each location with the global temperature distribution feature matrix, thereby determining whether there are anomalies at these locations and whether automatic activation of fire suppression equipment is necessary.

[0056] Specifically, the temperature time-series feature vectors of each location are used as query vectors, and matrix multiplication is performed with the global temperature distribution feature matrix to obtain multiple classification feature vectors. These classification feature vectors reflect the differences and similarities between each location and the global temperature distribution over time, thereby determining whether the location is currently in a normal state or a potential fire hazard state. It is worth noting that because temperature time-series data inherently possesses time-series properties, using matrix multiplication for processing better preserves the temperature change patterns at each location. Furthermore, considering the correlation and differences between locations over time, this method allows for a more accurate assessment of temperature changes at various locations within the mine, providing a more reliable and accurate basis for subsequent automated fire suppression decisions.

[0057] Next, the various classification feature vectors are processed by a classifier to obtain classification results. These results indicate whether to automatically activate the fire extinguishing equipment at each location. In other words, the classifier uses temperature changes at each location to promptly detect potential fire hazards and automatically trigger fire extinguishing equipment for preventative measures. Specifically, the calculated multiple classification feature vectors are used as input, and the classifier makes classification decisions to determine if there are any abnormalities in the temperature changes at each location. If the classification result for a location indicates a fire hazard, the system automatically activates the fire extinguishing equipment at that location to quickly and effectively eliminate the potential fire hazard. Meanwhile, for other locations in a normal state, monitoring and early warning systems continue to be implemented, improving the overall safety and stability of the mine.

[0058] Here, when using each temperature time-series feature vector as a query feature vector and calculating its matrix product with the global temperature distribution feature matrix to obtain the multiple classification feature vectors, since the global temperature distribution feature matrix represents the temperature time-series-spatial distribution correlation features of multiple locations in key parts of the mine, considering the statistical differences in temperature time-series distributions at different spatial locations and possible noise effects, the overall feature distribution of the global temperature distribution feature matrix has a low degree of regularity. Therefore, when mapping to the feature space of each temperature time-series feature vector through the query feature mapping of each temperature time-series feature vector, it leads to the irregularity of the feature distribution of the multiple classification feature vectors obtained, affecting the classification accuracy of the multiple classification feature vectors.

[0059] Based on this, the applicant of this application performs a second-order regularization of the Gaussian probability density parameter of the manifold surface on each of the plurality of classification feature vectors, for example, denoted as V, as follows:

[0060] v i ′=(μσ)v i 2 +v i μ+(v i -σ)μ 2

[0061] Where μ and σ are the eigenvalue set v i The mean and standard deviation of ∈V, and v i ′ is the feature value at the i-th position of the optimized classification feature vector V′.

[0062] Specifically, to address the problem of irregular distribution of high-dimensional features in the feature set of the classification feature vector V within the high-dimensional feature space of the query feature map, a second regularization of the feature values ​​is performed on the likelihood of the Gaussian probability density parameter of the class probability. This smooths the equidistantly distributed feature values ​​in the parameter space of the target Gaussian probability density parameter, thereby obtaining a regularized renormalization of the original probability density likelihood function of the high-dimensional feature manifold surface expression in the parameter space. This improves the classification accuracy of the optimized classification feature vector V′ through the classifier.

[0063] Based on this, this application proposes a distributed optical fiber temperature measurement and fire prevention control system for mines, comprising: a temperature data acquisition module for acquiring the discrete temperature time distribution of multiple locations in key parts of the mine collected by a distributed optical fiber temperature measurement system; a structuring module for arranging the discrete temperature time distribution of each location into multiple temperature time-series input vectors according to the time dimension; a temperature time-series feature extraction module for passing the multiple temperature time-series input vectors through a temperature time-series feature extractor containing a first convolutional layer and a second convolutional layer to obtain multiple temperature time-series feature vectors, wherein the first convolutional layer and the second convolutional layer respectively use one-dimensional convolutional kernels with different scales; and temperature time-series neighborhood association. The system includes a feature extraction module, which arranges the multiple temperature time-series feature vectors into a two-dimensional feature matrix and then uses a convolutional neural network model as a feature extractor to obtain a global temperature distribution feature matrix; a query module, which uses each temperature time-series feature vector as a query feature vector and calculates its matrix product with the global temperature distribution feature matrix to obtain multiple classification feature vectors; an optimization module, which performs regularization enhancement on each classification feature vector to obtain multiple optimized classification feature vectors; and a control result generation module, which passes each optimized classification feature vector through a classifier to obtain a classification result, wherein the classification result indicates whether to automatically activate the fire extinguishing equipment at each location.

[0064] After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0065] Exemplary System

[0066] Figure 1 This is a block diagram of a mine-use distributed fiber optic temperature measurement and fire prevention control system according to an embodiment of this application. Figure 1 As shown, the mine-use distributed optical fiber temperature measurement and fire prevention control system 300 according to an embodiment of this application includes: a temperature data acquisition module 310; a structuring module 320; a temperature time series feature extraction module 330; a temperature time series neighborhood association feature extraction module 340; a query module 350; an optimization module 360; and a control result generation module 370.

[0067] The temperature data acquisition module 310 is used to acquire the discrete temperature time distribution of multiple locations in key parts of the mine, collected by a distributed fiber optic temperature measurement system; the structuring module 320 is used to arrange the discrete temperature time distribution of each location into multiple temperature time-series input vectors according to the time dimension; the temperature time-series feature extraction module 330 is used to pass the multiple temperature time-series input vectors through a temperature time-series feature extractor containing a first convolutional layer and a second convolutional layer to obtain multiple temperature time-series feature vectors, wherein the first convolutional layer and the second convolutional layer use one-dimensional convolutional kernels with different scales; the temperature time-series neighborhood association feature extraction module 340 is used to extract the multiple temperature time-series feature vectors. The temperature time-series feature vectors are arranged into a two-dimensional feature matrix and then passed through a convolutional neural network model as a feature extractor to obtain a global temperature distribution feature matrix. The query module 350 is used to use each temperature time-series feature vector as a query feature vector and calculate the matrix product between it and the global temperature distribution feature matrix to obtain multiple classification feature vectors. The optimization module 360 ​​is used to perform regularization enhancement on each classification feature vector to obtain multiple optimized classification feature vectors. The control result generation module 370 is used to pass each optimized classification feature vector through a classifier to obtain a classification result, which is used to indicate whether to automatically activate the fire extinguishing equipment at each location.

[0068] Figure 2 This is a system architecture diagram of a mine-use distributed fiber optic temperature measurement and fire prevention control system according to an embodiment of this application. Figure 2As shown, in this network architecture, the temperature data acquisition module 310 first acquires the discrete temperature time distribution of multiple locations in key parts of the mine, collected by the distributed optical fiber temperature measurement system; then, the structuring module 320 arranges the discrete temperature time distribution of each location acquired by the temperature data acquisition module 310 into multiple temperature time-series input vectors according to the time dimension; the temperature time-series feature extraction module 330 passes the multiple temperature time-series input vectors obtained by the structuring module 320 through a temperature time-series feature extractor containing a first convolutional layer and a second convolutional layer to obtain multiple temperature time-series feature vectors, wherein the first convolutional layer and the second convolutional layer use one-dimensional convolutional kernels with different scales; then, the temperature time-series neighborhood association feature extraction module 340 extracts the temperature time-series feature vectors from the temperature time-series feature extraction module... After multiple temperature time-series feature vectors obtained by block 330 are arranged into a two-dimensional feature matrix, a convolutional neural network model, acting as a feature extractor, is used to obtain a global temperature distribution feature matrix. The query module 350 uses each temperature time-series feature vector obtained by the structuring module 320 as a query feature vector and calculates the matrix product between it and the global temperature distribution feature matrix obtained by the temperature time-series neighborhood association feature extraction module 340 to obtain multiple classification feature vectors. The optimization module 360 ​​performs regularization enhancement on each classification feature vector obtained by the query module 350 to obtain multiple optimized classification feature vectors. Then, the control result generation module 370 passes each optimized classification feature vector through a classifier to obtain a classification result, which is used to indicate whether the fire extinguishing equipment at each location is automatically activated.

[0069] Specifically, during the operation of the mine-use distributed fiber optic temperature measurement and fire prevention control system 300, the temperature data acquisition module 310 is used to acquire the discrete temporal distribution of temperature at multiple locations in key parts of the mine, collected by the distributed fiber optic temperature measurement system. It should be understood that mines contain a large number of flammable, explosive, and other hazardous materials, which will ignite and form a fire once stimulated by high temperatures or flames. Therefore, monitoring and early warning of temperature changes in various areas of the mine is necessary. In a specific example of this application, the distributed fiber optic temperature measurement system can acquire the discrete temporal distribution of temperature at multiple locations in key parts of the mine. In particular, the distributed fiber optic temperature measurement system can acquire time-series temperature data at various locations in the mine and transmit the data to a subsequent processing system for analysis and processing. Acquiring this time-series temperature distribution data allows for feature extraction and classification in subsequent technical solutions, thereby achieving the purpose of detecting temperature changes at various locations in the mine and automatically activating fire extinguishing equipment. Specifically, the basic principle of the DTS system is to use the ratio of Raman scattered light intensity generated by the interaction between the laser pulse and the fiber molecules to calculate the temperature value at different locations along the fiber. The location information is determined based on the propagation time of the light pulse in the fiber. In this way, the time-series temperature distribution of multiple locations (e.g., every 0.5 meters) of key parts of the mine (such as cables, conveyor belts, fans, etc.) can be obtained, that is, the temperature value of each location at different time points.

[0070] Specifically, during the operation of the mine-use distributed fiber optic temperature measurement and fire prevention control system 300, the structured module 320 is used to arrange the discrete temperature time-series distributions of each location into multiple temperature time-series input vectors according to the time dimension. This means that the data format of the discrete temperature time-series distributions of each location is adjusted to transform the original temperature data into an input format suitable for deep learning processing. Specifically, the temperature time-series distribution of each location is treated as a one-dimensional array, where each element represents the temperature value at a given time point. This results in multiple one-dimensional arrays of the same length (e.g., 1000), i.e., multiple temperature time-series input vectors.

[0071] Specifically, during the operation of the mine-use distributed fiber optic temperature measurement and fire prevention control system 300, the temperature time-series feature extraction module 330 is used to pass the multiple temperature time-series input vectors through a temperature time-series feature extractor containing a first convolutional layer and a second convolutional layer to obtain multiple temperature time-series feature vectors. The first and second convolutional layers each use one-dimensional convolutional kernels with different scales. That is, the multiple temperature time-series input vectors are passed through the temperature time-series feature extractor containing a first and second convolutional layer to obtain multiple temperature time-series feature vectors, where the first and second convolutional layers each use one-dimensional convolutional kernels with different scales. Here, the temperature time-series feature extractor performs multi-scale one-dimensional convolutional encoding on each temperature time-series input vector to capture the temperature time-series distribution features of multiple locations in key parts of the mine. The temperature time-series feature extractor includes: a first convolutional layer, a second convolutional layer parallel to the first convolutional layer, and a feature fusion layer connected to the first and second convolutional layers. The first convolutional layer uses a one-dimensional convolutional kernel of a first length, and the second convolutional layer uses a one-dimensional convolutional kernel of a second length. In the technical solution of this application, the temperature change patterns at different locations during a fire in a mine may vary significantly. Therefore, feature extraction and classification of temperature time-series data for different locations are necessary. Since temperature time-series data itself has time-series properties, convolutional neural networks can be used for feature extraction and classification. Using one-dimensional convolutional kernels of different scales for convolution operations can better adapt to temperature time-series data at different locations, thereby improving the accuracy of feature extraction and classification. Specifically, the first convolutional layer uses a smaller kernel to quickly capture local temperature change features and extract more refined local feature information; while the second convolutional layer uses a larger kernel to better capture more global temperature change features and extract more macroscopic global feature information. This allows local and global feature information to be organically combined to form a more complete and accurate temperature time series feature vector, providing a more reliable foundation for subsequent classification decisions.

[0072] Figure 3 This is a block diagram of the temperature time-series feature extraction module in a mine-use distributed fiber optic temperature measurement and fire prevention control system according to an embodiment of this application. Figure 3As shown, the temperature time-series feature extraction module 330 includes: a first temperature time-series feature extraction unit 331, used to input the temperature time-series input vector into the first convolutional layer of the temperature time-series feature extractor to obtain a first-scale temperature time-series feature vector, wherein the first convolutional layer has a first one-dimensional convolutional kernel of a first length; a second temperature time-series feature extraction unit 332, used to input the temperature time-series input vector into the second convolutional layer of the temperature time-series feature extractor to obtain a second-scale temperature time-series feature vector, wherein the second convolutional layer has a second one-dimensional convolutional kernel of a second length, the first length being different from the second length; and a multi-scale concatenation unit 333, used to concatenate the first-scale temperature time-series feature vector and the second-scale temperature time-series feature vector to obtain the temperature time-series feature vector. The first temperature time-series feature extraction unit 331 is used to: perform one-dimensional convolutional encoding on the temperature time-series input vector using the first convolutional layer of the temperature time-series feature extractor with the following one-dimensional convolution formula to obtain the first-scale temperature time-series feature vector; wherein the formula is:

[0073]

[0074] Where, a is the width of the first convolutional kernel in the x-direction, F(a) is the parameter vector of the first convolutional kernel, G(xa) is the local vector matrix operated with the convolutional kernel function, w is the size of the first convolutional kernel, X represents the temperature time-series input vector, and Cov(X) represents one-dimensional convolutional encoding of the temperature time-series input vector; and, the second temperature time-series feature extraction unit 332 is used to: use the second convolutional layer of the temperature time-series feature extractor to perform one-dimensional convolutional encoding on the temperature time-series input vector using the following one-dimensional convolution formula to obtain the second-scale temperature time-series feature vector; wherein, the formula is:

[0075]

[0076] Where b is the width of the second convolution kernel in the x direction, F(b) is the parameter vector of the second convolution kernel, G(xb) is the local vector matrix operated with the convolution kernel function, m is the size of the second convolution kernel, X represents the temperature time series input vector, and Cov(X) represents one-dimensional convolution encoding of the temperature time series input vector.

[0077] Specifically, during the operation of the mine-use distributed fiber optic temperature measurement and fire prevention control system 300, the temperature time-series neighborhood correlation feature extraction module 340 is used to arrange the multiple temperature time-series feature vectors into a two-dimensional feature matrix and then use a convolutional neural network model as a feature extractor to obtain a global temperature distribution feature matrix. That is, the temperature time-series distribution features of each location in the key parts of the mine are integrated into a global temperature distribution matrix, and a convolutional neural network model is used to perform local neighborhood feature extraction based on convolution kernels on the global temperature distribution matrix to capture the correlation pattern features between the temperature time-series distributions of each location in the key parts of the mine. Specifically, the temperature time-series feature vectors of each location are arranged into a two-dimensional feature matrix according to the time dimension, that is, each row represents the temperature time-series feature vector of a location at different time points. Then, this two-dimensional feature matrix is ​​input into the convolutional neural network model as a feature extractor for processing to obtain a feature matrix containing global temperature distribution information. Convolutional neural network (CNN) models can perform convolution and pooling operations on the input two-dimensional feature matrix to extract the correlations and patterns between different locations in a mine. In this way, the connections and differences between the temperature time-series feature vectors of various locations can be fully explored and integrated into a global temperature distribution feature matrix. This allows for a better understanding and grasp of temperature changes at various locations in the mine, providing a more comprehensive and accurate reference for subsequent classification decisions. In a specific example, the CNN includes multiple cascaded neural network layers, each comprising a convolutional layer, a pooling layer, and an activation layer. During the encoding process of the CNN, each layer performs kernel-based convolution processing on the input data during forward propagation, pools the convolutional feature map output by the convolutional layer using the pooling layer, and activates the pooled feature map output by the pooling layer using the activation layer.

[0078] Figure 4 This is a flowchart illustrating the convolutional neural network encoding in a mine-use distributed fiber optic temperature measurement and fire suppression control system according to an embodiment of this application. Figure 4As shown, the encoding process of the convolutional neural network includes: each layer of the convolutional neural network model, which serves as a feature extractor, performs the following operations on the input data during the forward propagation of the layer: S210, performing convolution processing on the input data to obtain a convolutional feature map; S220, performing pooling along the channel dimension on the convolutional feature map to obtain a pooled feature map; and S230, performing nonlinear activation on the pooled feature map to obtain an activation feature map; wherein, the output of the last layer of the convolutional neural network serving as a feature extractor is the global temperature distribution feature matrix, and the input of the first layer of the convolutional neural network serving as a feature extractor is a two-dimensional feature matrix formed by arranging the multiple temperature time-series feature vectors.

[0079] Specifically, during the operation of the mine-use distributed fiber optic temperature measurement and fire prevention control system 300, the query module 350 is used to calculate the matrix product between each temperature time-series feature vector and the global temperature distribution feature matrix to obtain multiple classification feature vectors, using each temperature time-series feature vector as a query feature vector. The temperature time-series feature vector at each location is used as a query vector, and matrix multiplication is performed with the global temperature distribution feature matrix to obtain multiple classification feature vectors. These classification feature vectors reflect the differences and similarities between each location and the global temperature distribution in the time dimension, thereby determining whether the location is currently in a normal state or a potential fire hazard state. It is worth mentioning that, since temperature time-series data itself has time-series properties, using matrix multiplication for processing can better preserve the temperature change patterns at each location and take into account the correlation and differences between locations in the time dimension.

[0080] Specifically, during the operation of the mine-use distributed optical fiber temperature measurement and fire prevention control system 300, the optimization module 360 ​​is used to perform regularization enhancement on each of the classification feature vectors to obtain multiple optimized classification feature vectors. Here, when using each temperature time-series feature vector as a query feature vector and calculating its matrix product with the global temperature distribution feature matrix to obtain the multiple classification feature vectors, since the global temperature distribution feature matrix represents the temperature time-series-spatial distribution correlation features of multiple locations in key parts of the mine, considering the statistical differences in temperature time-series distributions at different spatial locations and possible noise effects, the overall feature distribution regularization degree of the global temperature distribution feature matrix is ​​low. Therefore, when mapping to the feature space of each temperature time-series feature vector through the query feature mapping of each temperature time-series feature vector, it leads to the irregularity problem of the feature distribution of the obtained multiple classification feature vectors, affecting the classification accuracy of the multiple classification feature vectors. Based on this, the applicant of this application performs a second regularization of the Gaussian probability density parameter of the manifold surface on each of the multiple classification feature vectors, for example, denoted as V, specifically expressed as:

[0081] v i ′=(μσ)v i 2 +v i μ+(v i -σ)μ 2

[0082] Where μ and σ are the mean and standard deviation of the feature values ​​at each position of the classification feature vector, and v i ′ is the eigenvalue at the i-th position of the optimized classification feature vector. Specifically, to address the problem of irregular distribution of high-dimensional features in the feature set of the classification feature vector V within the high-dimensional feature space of the query feature map, a second regularization of the eigenvalues ​​is performed on the likelihood of the Gaussian probability density parameter of the class probability. This smooths the equidistantly distributed eigenvalues ​​within the parameter space of the target Gaussian probability density parameter, thereby obtaining a regularized renormalization of the original probability density likelihood function of the high-dimensional feature manifold surface expression within the parameter space. This improves the classification accuracy of the optimized classification feature vector V′ by the classifier.

[0083] Specifically, during the operation of the mine-use distributed fiber optic temperature measurement and fire suppression control system 300, the control result generation module 370 is used to pass the optimized classification feature vectors through a classifier to obtain classification results. These classification results indicate whether the fire suppression equipment at each location is automatically activated. That is, in the technical solution of this application, after obtaining the optimized classification feature vectors, they are further passed through a classifier to obtain classification results indicating whether the fire suppression equipment at each location is automatically activated. Specifically, the classifier processes the optimized classification feature vectors using the following formula to obtain the classification results, where the formula is:

[0084] O = softmax{)W n B n ):…:)W1,B1)|X}, where W1 to W n The weight matrix is ​​B1 to B1. n Let X be the bias vector and X be the optimized classification feature vector. Specifically, the classifier includes multiple fully connected layers and a Softmax layer cascaded with the last fully connected layer of the multiple fully connected layers. In the classification process of the classifier, the optimized classification feature vector is fully encoded multiple times using the multiple fully connected layers of the classifier to obtain an encoded classification feature vector. Then, the encoded classification feature vector is input into the Softmax layer of the classifier, that is, the Softmax classification function is used to classify the encoded classification feature vector to obtain a classification label. In other words, the classifier uses temperature changes at various locations to promptly detect potential fire hazards and automatically trigger fire extinguishing equipment for preventative measures. Specifically, the calculated multiple classification feature vectors are used as input, and the classifier makes classification decisions to determine whether there are abnormal temperature changes at various locations. If the classification result of a location indicates a fire hazard, the system automatically activates the fire extinguishing equipment at that location to quickly and effectively eliminate the potential fire hazard. Meanwhile, for other locations in a normal state, monitoring and early warning can continue, improving the overall safety and stability of the mine.

[0085] In summary, the distributed optical fiber temperature measurement and fire prevention control system 300 for mines according to the embodiments of this application is explained. It uses a deep learning neural network model to mine the temperature time-series distribution correlation characteristics of various locations in key parts of the mine and adaptively controls the fire extinguishing equipment at each location. In this way, it can better understand and grasp the temperature changes at various locations in the mine, and thus provide a more reliable and accurate basis for subsequent automated fire extinguishing decisions.

[0086] As described above, the mine-use distributed fiber optic temperature measurement and fire prevention control system according to the embodiments of this application can be implemented in various terminal devices. In one example, the mine-use distributed fiber optic temperature measurement and fire prevention control system 300 according to the embodiments of this application can be integrated into the terminal device as a software module and / or hardware module. For example, the mine-use distributed fiber optic temperature measurement and fire prevention control system 300 can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the mine-use distributed fiber optic temperature measurement and fire prevention control system 300 can also be one of many hardware modules of the terminal device.

[0087] Alternatively, in another example, the mine-use distributed fiber optic temperature measurement and fire prevention control system 300 and the terminal device can also be separate devices, and the mine-use distributed fiber optic temperature measurement and fire prevention control system 300 can be connected to the terminal device via wired and / or wireless networks, and transmit interactive information in accordance with the agreed data format.

[0088] Exemplary methods

[0089] Figure 5 This is a flowchart of a mine-use distributed optical fiber temperature measurement and fire prevention control method according to an embodiment of this application. Figure 5As shown, the mine-use distributed optical fiber temperature measurement fire prevention and extinguishing control method according to an embodiment of this application includes the following steps: S110, acquiring the discrete temperature time distribution of multiple locations in key parts of the mine collected by the distributed optical fiber temperature measurement system; S120, arranging the discrete temperature time distribution of each location into multiple temperature time-series input vectors according to the time dimension; S130, passing the multiple temperature time-series input vectors through a temperature time-series feature extractor containing a first convolutional layer and a second convolutional layer to obtain multiple temperature time-series feature vectors, wherein the first convolutional layer and the second convolutional layer respectively use one-dimensional convolutional kernels with different scales; S140, After arranging the multiple temperature time-series feature vectors into a two-dimensional feature matrix, a convolutional neural network model, acting as a feature extractor, is used to obtain a global temperature distribution feature matrix; S150, each temperature time-series feature vector is used as a query feature vector, and its matrix product with the global temperature distribution feature matrix is ​​calculated to obtain multiple classification feature vectors; S160, each classification feature vector is subjected to regularization enhancement to obtain multiple optimized classification feature vectors; and S170, each optimized classification feature vector is passed through a classifier to obtain a classification result, the classification result being used to indicate whether the fire extinguishing equipment at each location is automatically activated. In one example, in the above-mentioned distributed optical fiber temperature measurement and fire prevention control method for mines, step S130 includes: inputting the temperature time-series input vector into the first convolutional layer of the temperature time-series feature extractor to obtain a first-scale temperature time-series feature vector, wherein the first convolutional layer has a first one-dimensional convolutional kernel of a first length; inputting the temperature time-series input vector into the second convolutional layer of the temperature time-series feature extractor to obtain a second-scale temperature time-series feature vector, wherein the second convolutional layer has a second one-dimensional convolutional kernel of a second length, the first length being different from the second length; and concatenating the first-scale temperature time-series feature vector and the second-scale temperature time-series feature vector to obtain the temperature time-series feature vector. The first convolutional layer and the second convolutional layer are parallel, and the temperature time-series feature extractor further includes a multi-scale feature fusion layer connected to both the first convolutional layer and the second convolutional layer. The process of inputting the temperature time-series input vector into the first convolutional layer of the temperature time-series feature extractor to obtain a first-scale temperature time-series feature vector includes: using the first convolutional layer of the temperature time-series feature extractor to perform one-dimensional convolutional encoding on the temperature time-series input vector using the following one-dimensional convolution formula to obtain the first-scale temperature time-series feature vector; wherein, the formula is:

[0090]

[0091] Where, a is the width of the first convolutional kernel in the x-direction, F(a) is the parameter vector of the first convolutional kernel, G(xa) is the local vector matrix operated with the convolutional kernel function, w is the size of the first convolutional kernel, X represents the temperature time-series input vector, and Cov(X) represents one-dimensional convolutional encoding of the temperature time-series input vector; and, inputting the temperature time-series input vector into the second convolutional layer of the temperature time-series feature extractor to obtain the second-scale temperature time-series feature vector includes: using the second convolutional layer of the temperature time-series feature extractor to perform one-dimensional convolutional encoding on the temperature time-series input vector using the following one-dimensional convolution formula to obtain the second-scale temperature time-series feature vector; wherein, the formula is:

[0092]

[0093] Where b is the width of the second convolution kernel in the x direction, F(b) is the parameter vector of the second convolution kernel, G(xb) is the local vector matrix operated with the convolution kernel function, m is the size of the second convolution kernel, X represents the temperature time series input vector, and Cov(X) represents one-dimensional convolution encoding of the temperature time series input vector.

[0094] In one example, in the above-mentioned distributed optical fiber temperature measurement and fire prevention control method for mines, step S140 includes: using each layer of the convolutional neural network model as a feature extractor to process the input data in the forward propagation of the layer, performing convolution processing on the input data to obtain a convolutional feature map; performing pooling along the channel dimension on the convolutional feature map to obtain a pooled feature map; and performing nonlinear activation on the pooled feature map to obtain an activation feature map; wherein, the output of the last layer of the convolutional neural network as a feature extractor is the global temperature distribution feature matrix, and the input of the first layer of the convolutional neural network as a feature extractor is a two-dimensional feature matrix formed by arranging the multiple temperature time-series feature vectors. The convolutional neural network model as a feature extractor is a deep residual network model.

[0095] In one example, in the above-mentioned distributed optical fiber temperature measurement and fire prevention control method for mines, step S160 includes: regularizing and strengthening each classification feature vector using the following optimization formula to obtain multiple optimized classification feature vectors; wherein, the formula is:

[0096] v i ′=(μσ)v i 2 +v i μ+(v i -σ)μ 2

[0097] Where μ and σ are the mean and standard deviation of the feature values ​​at each position of the classification feature vector, and v i ′ is the feature value at the i-th position of the optimized classification feature vector.

[0098] In one example, in the above-mentioned distributed optical fiber temperature measurement and fire prevention control method for mines, step S170 includes: using multiple fully connected layers of the classifier to perform fully connected encoding on the optimized classification feature vector to obtain an encoded classification feature vector; and passing the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

[0099] In summary, the distributed optical fiber temperature measurement and fire prevention control method for mines according to the embodiments of this application is explained. It uses a deep learning neural network model to mine the temperature time-series distribution correlation characteristics of various locations in key parts of the mine and uses this to adaptively control the fire extinguishing equipment at each location. In this way, it is possible to better understand and grasp the temperature changes at various locations in the mine, thereby providing a more reliable and accurate basis for subsequent automated fire extinguishing decisions.

[0100] Exemplary electronic devices

[0101] Below, for reference Figure 6 This describes an electronic device according to embodiments of the present application.

[0102] Figure 6 A block diagram of an electronic device according to an embodiment of this application is illustrated.

[0103] like Figure 6 As shown, the electronic device 10 includes one or more processors 11 and memory 12.

[0104] The processor 11 may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.

[0105] The memory 12 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may execute the program instructions to implement the functions and / or other desired functions in the mine distributed fiber optic temperature measurement and fire prevention control system of the various embodiments of this application described above. Various contents, such as multi-scale feature vectors of power generation, may also be stored in the computer-readable storage medium.

[0106] In one example, the electronic device 10 may also include an input device 13 and an output device 14, which are interconnected via a bus system and / or other forms of connection mechanism (not shown).

[0107] The input device 13 may include, for example, a keyboard, a mouse, etc.

[0108] The output device 14 can output various information to the outside, including classification results. The output device 14 may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, etc.

[0109] Of course, for the sake of simplicity, Figure 6 Only some of the components of the electronic device 10 relevant to this application are shown in this illustration; components such as buses, input / output interfaces, etc., are omitted. In addition, the electronic device 10 may include any other suitable components depending on the specific application.

[0110] Exemplary computer program products and computer-readable storage media

[0111] In addition to the methods and devices described above, embodiments of this application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps of the functions in the mining distributed fiber optic temperature measurement and fire extinguishing control method according to various embodiments of this application as described in the "Exemplary Systems" section of this specification.

[0112] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0113] Furthermore, embodiments of this application may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps of the integrated energy storage management method according to various embodiments of this application as described in the "Exemplary Systems" section above.

[0114] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0115] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0116] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0117] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0118] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0119] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A mine distributed optical fiber temperature measurement fire prevention and extinguishing control system, characterized in that, The method comprises the following steps: a temperature data acquisition module for acquiring temperature time series discrete distribution of multiple positions of key parts of a mine collected by a distributed optical fiber temperature measurement system; a structured module for arranging the temperature time series discrete distribution of each position into a plurality of temperature time series input vectors according to the time dimension respectively; a temperature time series feature extraction module for obtaining a plurality of temperature time series feature vectors by passing the plurality of temperature time series input vectors through a temperature time series feature extractor comprising a first convolutional layer and a second convolutional layer respectively, wherein the first convolutional layer and the second convolutional layer use one-dimensional convolution kernels with different scales respectively; a temperature time series neighborhood correlation feature extraction module for arranging the plurality of temperature time series feature vectors into a two-dimensional feature matrix and then passing the two-dimensional feature matrix through a convolutional neural network model as a feature extractor to obtain a temperature global distribution feature matrix; a query module for taking each temperature time series feature vector as a query feature vector, calculating the matrix product between the query feature vector and the temperature global distribution feature matrix to obtain a plurality of classification feature vectors; an optimization module for respectively regularizing and strengthening each classification feature vector to obtain a plurality of optimized classification feature vectors; and a control result generation module for passing each optimized classification feature vector through a classifier to obtain a classification result, wherein the classification result is used to indicate whether to automatically start a fire extinguishing device at each position.

2. The distributed optical fiber temperature measurement fire control system for mine of claim 1, wherein, The first convolutional layer and the second convolutional layer are parallel, and the temperature time series feature extractor further comprises a multi-scale feature fusion layer connected to the first convolutional layer and the second convolutional layer simultaneously.

3. The distributed optical fiber temperature measurement fire control system for mine of claim 2, characterized in that, The temperature time series feature extraction module comprises: a first temperature time series feature extraction unit for inputting the temperature time series input vector into the first convolutional layer of the temperature time series feature extractor to obtain a first scale temperature time series feature vector, wherein the first convolutional layer has a first one-dimensional convolution kernel with a first length; a second temperature time series feature extraction unit for inputting the temperature time series input vector into the second convolutional layer of the temperature time series feature extractor to obtain a second scale temperature time series feature vector, wherein the second convolutional layer has a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and a multi-scale concatenation unit for concatenating the first scale temperature time series feature vector and the second scale temperature time series feature vector to obtain the temperature time series feature vector. The first temperature time series feature extraction unit is configured to use the first convolutional layer of the temperature time series feature extractor to perform one-dimensional convolution coding on the temperature time series input vector according to the following one-dimensional convolution formula to obtain the first scale temperature time series feature vector. The formula is as follows: wherein a is the width of the first convolution kernel in the x direction, F(a) is the first convolution kernel parameter vector, G(x-a) is the local vector matrix operated with the convolution kernel function, w is the size of the first one-dimensional convolution kernel, X represents the temperature time series input vector, and Cov(X) represents one-dimensional convolution coding on the temperature time series input vector; and The second temperature time sequence feature extraction unit is configured to use a second convolutional layer of the temperature time sequence feature extractor to perform one-dimensional convolution coding on the temperature time sequence input vector according to a one-dimensional convolution formula to obtain the second scale temperature time sequence feature vector. The formula is as follows: wherein b is the width of the second convolution kernel in the x direction, F(b) is a second convolution kernel parameter vector, G(x-b) is a local vector matrix for convolution kernel function operation, m is the size of the second one-dimensional convolution kernel, X represents the temperature time sequence input vector, and Cov(X) represents one-dimensional convolution coding on the temperature time sequence input vector.

4. The distributed optical fiber temperature measurement fire control system for mine of claim 3, characterized in that, The temperature time sequence neighborhood correlation feature extraction module is configured to use each layer of the convolutional neural network model serving as a feature extractor to respectively perform the following operations on input data in the forward transmission of the layer: perform convolution processing on the input data to obtain a convolution feature map; perform channel dimension pooling on the convolution feature map to obtain a pooling feature map; and perform nonlinear activation on the pooling feature map to obtain an activated feature map. The output of the last layer of the convolutional neural network serving as a feature extractor is the temperature global distribution feature matrix, and the input of the first layer of the convolutional neural network serving as a feature extractor is a two-dimensional feature matrix formed by arranging the plurality of temperature time sequence feature vectors.

5. The distributed optical fiber temperature measurement fire control system for mine of claim 4, characterized in that, The convolutional neural network model serving as a feature extractor is a deep residual network model.

6. The distributed optical fiber temperature measurement fire control system for mine of claim 5, characterized in that, The optimization module is configured to regularize and strengthen each classification feature vector according to an optimization formula to obtain a plurality of optimized classification feature vectors. The formula is as follows: v′ i = (μσ)v i 2 +v i μ+(v i -σ)μ 2 where μ and σ are the mean and standard deviation of the feature values of the classification feature vector at the respective positions, and v' i is the feature value of the i-th position of the optimized classification feature vector.

7. The distributed optical fiber temperature measurement fire control system for mine of claim 6, characterized in that, The control result generation module includes: a fully connected coding unit configured to use a plurality of fully connected layers of the classifier to perform fully connected coding on the optimized classification feature vector to obtain a coded classification feature vector; and a classification result generation unit configured to pass the coded classification feature vector through a Softmax classification function of the classifier to obtain the classification result.

Images