Intelligent mine disaster early warning device and system

Abstract

The invention discloses an intelligent mine disaster early warning device and system, and the device comprises a multi-mode sensing array which comprises a bionic mine pressure sensor group, a gas spectrum fingerprint recognition module, and a micro-seismic beam forming array; the adaptive communication network comprises an edge preprocessing unit and a dual-mode communication module; and the dynamic risk assessment server is configured with a spatial-temporal feature fusion neural network and a threshold evolution engine. According to the invention, three-dimensional reconstruction of a stress field is realized through a tree-shaped groove substrate of the bionic mine pressure sensor group and a three-dimensional piezoelectric network, and a multi-mode sensing system covering stress, gas and slight shock is formed in combination with wide-spectrum gas fingerprint identification of the micro Fourier infrared spectrometer and the electrochemical sensor group. The spatial resolution and the gas type recognition coverage rate can be remarkably improved, dynamic correlation analysis of cross-modal data is achieved through the spatio-temporal feature fusion neural network, the disaster precursor feature capture sensitivity can be effectively improved, and the false alarm rate can be effectively reduced.

Description

Technical Field

[0001] This invention relates to the field of mine disaster early warning system technology, and in particular to an intelligent mine disaster early warning device and system. Background Technology

[0002] Current mine disaster early warning systems mainly monitor parameters such as underground stress, gas concentration, and microseismic activity in real time by deploying sensor networks, and combine them with threshold alarm mechanisms to achieve disaster early warning, playing an important role in preventing accidents such as roof falls, gas outbursts, and water inrushes.

[0003] However, traditional systems often use a single type of sensor to work independently. Stress monitoring relies on single-point pressure gauges, gas detection is limited to a single technical path of electrochemistry or infrared spectroscopy, and the accuracy of microseismic positioning is limited by the deployment method of planar arrays. This results in the incomplete capture of key precursor features such as stress field gradient changes, trace release of harmful gases, and spatial evolution of seismic sources, making it difficult to establish a multi-physics correlation analysis model.

[0004] Furthermore, existing sensors are prone to data distortion under conditions of strong electromagnetic interference and high humidity. The communication network lacks a dynamic anti-interference mechanism, and the fixed early warning threshold cannot adapt to the dynamic changes in underground geological conditions. False alarms or missed alarms often lead to unnecessary production stoppages or delays in handling the situation.

[0005] To address the above issues, we have developed an intelligent mine disaster early warning device and system. Summary of the Invention

[0006] This invention discloses an intelligent mine disaster early warning device and system, which aims to solve the technical problems in the background art.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] An intelligent mine disaster early warning system includes:

[0009] The multimodal sensing array includes a biomimetic mine pressure sensor group, a gas spectral fingerprint recognition module, and a micro-vibration beamforming array;

[0010] An adaptive communication network, including an edge preprocessing unit and a dual-mode communication module;

[0011] Dynamic risk assessment server, configured with spatiotemporal feature fusion neural network and threshold evolution engine;

[0012] A tiered response mechanism, including emergency control interfaces and visual early warning terminals;

[0013] The biomimetic mine pressure sensor group uses piezoelectric ceramic units arranged in an array, and the micro-vibration beamforming array consists of a three-dimensional monitoring network composed of multiple MEMS accelerometers.

[0014] In a preferred embodiment, the biomimetic mine pressure sensor group specifically includes:

[0015] The substrate has a grooved structure etched on its surface;

[0016] Five piezoelectric ceramic units are embedded at the intersection of the grooves;

[0017] A stress-transmitting layer is applied to the surface of the piezoelectric ceramic unit.

[0018] In a preferred embodiment, the gas spectral fingerprint recognition module includes:

[0019] Miniature Fourier transform infrared spectrometer, operating band covering 2-14μm;

[0020] The electrochemical sensor array includes an 8-channel detection unit for CO, CH4, H2S, O2, NO2, SO2, NH3, and Cl2.

[0021] The gas path switching device uses a microfluidic chip to achieve time-sequential control and sampling of multiple gas channels.

[0022] In a preferred embodiment, the adaptive communication network comprises:

[0023] The environmental sensing unit monitors parameters such as temperature, humidity, and electromagnetic interference intensity in real time.

[0024] The routing decision module automatically switches to industrial Ethernet communication mode when the electromagnetic interference intensity exceeds a threshold.

[0025] The data compression unit employs a time-frequency domain hybrid compression algorithm based on wavelet transform.

[0026] In a preferred embodiment, the spatiotemporal feature fusion neural network comprises:

[0027] A three-dimensional convolutional subnetwork with 5×5×5 3D convolutional kernels;

[0028] Temporal attention subnetwork, calculates feature association weights at each time step;

[0029] The feature fusion layer uses a channel attention mechanism to dynamically allocate the fusion ratio of spatial and temporal features.

[0030] In a preferred embodiment, the threshold evolution engine performs the following steps:

[0031] S1. Collect environmental baseline data from the most recent N hours to establish a reference distribution;

[0032] S2. Calculate the Wasserstein distance between the current monitoring data and the reference distribution;

[0033] S3. Threshold optimization is triggered when the distance exceeds the historical maximum value;

[0034] S4. Determine new threshold combinations through Monte Carlo tree search.

[0035] In a preferred embodiment, the hierarchical response actuator includes:

[0036] Activate the data review process when a blue alert is issued;

[0037] When a yellow alert is issued, enhanced ventilation will be activated in certain areas.

[0038] When an orange alert is issued, implement personnel evacuation route planning;

[0039] A red alert triggers a full mine power outage and emergency drainage.

[0040] An intelligent mine disaster early warning device includes:

[0041] The explosion-proof housing has a double-layer structure consisting of a stainless steel substrate and a ceramic coating.

[0042] A multimodal sensor array, embedded in the front face of the housing, includes:

[0043] Bionic mining pressure detection unit;

[0044] Gas analysis channel;

[0045] Microseismic sensing module;

[0046] The edge processing module is equipped with a pruned and optimized 3D convolutional neural network model.

[0047] A hybrid communication interface, including an exposed Ethernet port and a built-in LoRa antenna array.

[0048] The intelligent mine disaster early warning device and system provided by this invention has the following advantages:

[0049] 1. By using a tree-shaped grooved substrate and a three-dimensional piezoelectric network to reconstruct the stress field in a biomimetic mine pressure sensor array, and combining it with a micro Fourier transform infrared spectrometer and an electrochemical sensor array for broadband gas fingerprint recognition, a multimodal sensing system covering stress, gas, and micro-seismic activity is formed. This can significantly improve spatial resolution and gas type recognition coverage. Furthermore, by using a spatiotemporal feature fusion neural network to achieve dynamic correlation analysis of cross-modal data, the sensitivity of disaster precursor feature capture can be effectively improved and the false alarm rate can be reduced.

[0050] 2. A dynamic optimization mechanism is constructed using an environment-aware dual-mode communication network and a threshold evolution engine to ensure communication reliability under strong electromagnetic interference. Monte Carlo tree search is then used to optimize early warning thresholds in real time. Tiered response is achieved through digital twin path planning and equipment linkage control, shortening disaster response time and improving emergency decision-making efficiency. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the overall structure of an intelligent mine disaster early warning system proposed in this invention.

[0052] Figure 2 This is a schematic diagram of the gas spectral fingerprint recognition module structure of an intelligent mine disaster early warning device and system proposed in this invention.

[0053] Figure 3 This is a schematic diagram of the spatiotemporal feature fusion neural network structure of an intelligent mine disaster early warning device and system proposed in this invention.

[0054] Figure 4 This is a schematic diagram illustrating the execution steps of the threshold evolution engine in an intelligent mine disaster early warning system proposed in this invention.

[0055] Figure 5 This is a schematic diagram of the structure of an intelligent mine disaster early warning device proposed in this invention.

[0056] In the attached diagram: 1. Explosion-proof housing; 2. Multimodal sensor array; 201. Bionic mine pressure detection unit; 202. Gas analysis channel; 203. Micro-vibration sensing module; 204. Edge processing module; 205. Hybrid communication interface; 2051. Ethernet port; 2052. LoRa antenna array. Detailed Implementation

[0057] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and marked in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0058] This invention discloses an intelligent mine disaster early warning device and system.

[0059] Reference Figure 1 , Figure 2, Figure 3 , Figure 4 and Figure 5 As shown, an intelligent mine disaster early warning system includes:

[0060] The multimodal sensing array includes a biomimetic mine pressure sensor group, a gas spectral fingerprint recognition module, and a micro-vibration beamforming array;

[0061] The biomimetic mine pressure sensor group specifically includes:

[0062] The substrate has a grooved structure etched on its surface;

[0063] Five piezoelectric ceramic units are embedded at the intersection of the grooves;

[0064] A stress-transferring layer covers the surface of the piezoelectric ceramic unit;

[0065] The gas spectral fingerprint recognition module includes:

[0066] Miniature Fourier transform infrared spectrometer, operating band covering 2-14μm;

[0067] The electrochemical sensor array includes an 8-channel detection unit for CO, CH4, H2S, O2, NO2, SO2, NH2, and Cl2.

[0068] The gas path switching device uses a microfluidic chip to achieve time-sequential control and sampling of multiple gas channels;

[0069] The main body of the device employs a multimodal sensing array for collaborative monitoring. The biomimetic mine pressure sensor group utilizes a trench structure etched onto the substrate surface, embedding five piezoelectric ceramic units at the trench intersections to form a stress transmission network. A stress-conducting layer made of polyurethane elastomer is then coated onto the surface of the piezoelectric units, enabling three-dimensional sensing of mine pressure distribution. The gas detection module integrates a miniature Fourier transform infrared spectrometer and an 8-channel electrochemical sensor group. The spectrometer is equipped with a germanium-based waveguide to achieve a 2-14μm wideband scanning. Combined with a time-switching gas path constructed using a microfluidic chip, multiple gas samples are alternately introduced through a periodically rotating sampling disk.

[0070] An adaptive communication network, including an edge preprocessing unit and a dual-mode communication module;

[0071] The adaptive communication network further includes:

[0072] The environmental sensing unit monitors parameters such as temperature, humidity, and electromagnetic interference intensity in real time.

[0073] The routing decision module automatically switches to industrial Ethernet communication mode when the electromagnetic interference intensity exceeds a threshold.

[0074] The data compression unit employs a time-frequency domain hybrid compression algorithm based on wavelet transform;

[0075] Dynamic risk assessment server, configured with spatiotemporal feature fusion neural network and threshold evolution engine;

[0076] The dynamic risk assessment server deploys a spatiotemporal feature fusion neural network. A 3D convolutional sub-network uses a 5×5×5 kernel to extract tunnel structural features, while a temporal attention sub-network calculates the correlation weights at each time step. The spatiotemporal feature fusion ratio is dynamically adjusted through a channel attention mechanism. A threshold evolution engine continuously collects environmental baseline data to construct a Gaussian mixture model. When the Wasserstein distance between the real-time monitoring data and the baseline distribution exceeds a historical extreme, a Monte Carlo tree search algorithm is triggered to find the optimal threshold combination in the 3D parameter space.

[0077] The spatiotemporal feature fusion neural network includes:

[0078] A three-dimensional convolutional subnetwork with 5×5×5 3D convolutional kernels;

[0079] Temporal attention subnetwork, calculates feature association weights at each time step;

[0080] The feature fusion layer uses a channel attention mechanism to dynamically allocate the fusion ratio of spatial and temporal features;

[0081] The threshold evolution engine performs the following steps:

[0082] S1. Collect environmental baseline data from the most recent N hours to establish a reference distribution;

[0083] S2. Calculate the Wasserstein distance between the current monitoring data and the reference distribution;

[0084] S3. Threshold optimization is triggered when the distance exceeds the historical maximum value;

[0085] S4. Determine new threshold combinations through Monte Carlo tree search;

[0086] A tiered response mechanism, including emergency control interfaces and visual early warning terminals;

[0087] The biomimetic mine pressure sensor group uses piezoelectric ceramic units arranged in an array, and the micro-vibration beamforming array consists of a three-dimensional monitoring network composed of multiple MEMS accelerometers.

[0088] The three-dimensional monitoring network consists of a micro-seismic beamforming array constructed from 24 MEMS accelerometers. Each node is deployed on the working surface in an icosahedral configuration, and vibration source localization is achieved through a phase synchronization algorithm. When the acquired data is transmitted through the adaptive communication network, the edge preprocessing unit uses a wavelet transform algorithm for time-frequency hybrid compression, and the environmental sensing unit monitors the electromagnetic field strength in real time. When the interference intensity exceeds 50 dBμV / m, the routing decision module automatically switches the wireless mesh network to industrial Ethernet transmission mode.

[0089] The hierarchical response execution mechanism includes:

[0090] Activate the data review process when a blue alert is issued;

[0091] When a yellow alert is issued, enhanced ventilation will be activated in certain areas.

[0092] When an orange alert is issued, implement personnel evacuation route planning;

[0093] A red alert triggers a full mine power outage and emergency drainage.

[0094] The tiered response mechanism establishes a multi-level linkage system: when a blue alert is triggered, a data cross-validation process is initiated; a yellow alert activates the frequency conversion control of local ventilation fans; an orange alert uses a digital twin model to plan the optimal evacuation route; and a red alert links the central control system to execute a mine-wide power outage and activate high-power drainage pump units. The entire system achieves multi-dimensional early warning and tiered response to mine disasters through multi-source information fusion and intelligent decision-making algorithms.

[0095] An intelligent mine disaster early warning device includes:

[0096] The explosion-proof housing 1 has a double-layer structure consisting of a stainless steel substrate and a ceramic coating;

[0097] The multimodal sensor array 2 is embedded in the front end face of the housing and includes:

[0098] Bionic mining pressure detection unit 201;

[0099] Gas analysis channel 202;

[0100] Microseismic sensing module 203;

[0101] Edge processing module 204, on which a pruned and optimized three-dimensional convolutional neural network model is mounted;

[0102] The hybrid communication interface 205 includes an exposed Ethernet port 2051 and a built-in LoRa antenna array 2052.

[0103] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. The substitutions may be replacements of some structures, devices, or method steps, or they may be complete technical solutions. Equivalent substitutions or modifications made to the technical solutions and inventive concepts of the present invention should all be covered within the scope of protection of the present invention.

Claims

1. An intelligent mine disaster early warning system, characterized in that, include: The multimodal sensing array includes a biomimetic mine pressure sensor group, a gas spectral fingerprint recognition module, and a micro-vibration beamforming array; An adaptive communication network, including an edge preprocessing unit and a dual-mode communication module; Dynamic risk assessment server, configured with spatiotemporal feature fusion neural network and threshold evolution engine; A tiered response mechanism, including emergency control interfaces and visual early warning terminals; The biomimetic mine pressure sensor group uses piezoelectric ceramic units arranged in an array, and the micro-vibration beamforming array consists of a three-dimensional monitoring network composed of multiple MEMS accelerometers.

2. The intelligent mine disaster early warning system according to claim 1, characterized in that, The biomimetic mine pressure sensor group specifically includes: The substrate has a grooved structure etched on its surface; Five piezoelectric ceramic units are embedded at the intersection of the grooves; A stress-transmitting layer is applied to the surface of the piezoelectric ceramic unit.

3. The intelligent mine disaster early warning system according to claim 1, characterized in that, The gas spectral fingerprint recognition module includes: Miniature Fourier transform infrared spectrometer, operating band covering 2-14μm; The electrochemical sensor array includes an 8-channel detection unit for CO, CH4, H2S, O2, NO2, SO2, NH3, and Cl2. The gas path switching device uses a microfluidic chip to achieve time-sequential control and sampling of multiple gas channels.

4. The intelligent mine disaster early warning system according to claim 1, characterized in that, The adaptive communication network further includes: The environmental sensing unit monitors parameters such as temperature, humidity, and electromagnetic interference intensity in real time. The routing decision module automatically switches to industrial Ethernet communication mode when the electromagnetic interference intensity exceeds a threshold. The data compression unit employs a time-frequency domain hybrid compression algorithm based on wavelet transform.

5. The intelligent mine disaster early warning system according to claim 1, characterized in that, The spatiotemporal feature fusion neural network includes: A three-dimensional convolutional subnetwork with 5×5×5 3D convolutional kernels; Temporal attention subnetwork, calculates feature association weights at each time step; The feature fusion layer uses a channel attention mechanism to dynamically allocate the fusion ratio of spatial and temporal features.

6. The intelligent mine disaster early warning system according to claim 5, characterized in that, The threshold evolution engine performs the following steps: S1. Collect environmental baseline data from the most recent N hours to establish a reference distribution; S2. Calculate the Wasserstein distance between the current monitoring data and the reference distribution; S3. Threshold optimization is triggered when the distance exceeds the historical maximum value; S4. Determine new threshold combinations through Monte Carlo tree search.

7. The intelligent mine disaster early warning system according to claim 1, characterized in that, The hierarchical response execution mechanism includes: Activate the data review process when a blue alert is issued; When a yellow alert is issued, enhanced ventilation will be activated in certain areas. When an orange alert is issued, implement personnel evacuation route planning; A red alert triggers a full mine power outage and emergency drainage.

8. An intelligent mine disaster early warning device, characterized in that, include: The explosion-proof housing (1) has a double-layer structure consisting of a stainless steel substrate and a ceramic coating; A multimodal sensor array (2) is embedded in the front end face of the housing, and includes: Bionic mine pressure detection unit (201); Gas analysis channel (202); Microseismic sensing module (203); An edge processing module (204) is equipped with a pruned and optimized three-dimensional convolutional neural network model; The hybrid communication interface (205) includes an exposed Ethernet port (2051) and a built-in LoRa antenna array (2052).

Images