Mine safety trend prediction data communication and visualization method and system

Abstract

The application relates to the technical field of mine safety early warning, in particular to a mine safety trend prediction data communication and visualization method and system, which comprises the following steps: calculating a trend approximation rate; when the trend approximation rate is greater than a preset threshold corresponding to a real-time physical quantity type, switching to a preset second frequency by a measurement node to acquire the real-time physical quantity and generate an alarm message; identifying a network identifier, closing a sending door of a preset conventional queue to cut off conventional data, establishing an exclusive channel to transmit the alarm message; reducing the rendering precision of a three-dimensional model of an area outside a target coordinate based on a redirection bit by a visualization terminal, performing three-dimensional rendering on a warning area according to a rendering refresh priority mapped by a peak time; and synchronously receiving the alarm message by an explosion-proof execution device through the exclusive channel, and cutting off the power supply of a power supply partition or a feeder loop corresponding to the target coordinate. The application can solve the control delay caused by message queuing and rendering stagnation, and realize rapid physical power-off.

Description

Technical Field

[0001] This application relates to the technical field of mine safety early warning, and in particular to a method and system for data communication and visualization of mine safety trend prediction. Background Technology

[0002] Modern mines typically deploy numerous multi-physical quantity sensors in deep fully mechanized mining faces and key roadways to acquire real-time safety status parameters such as gas concentration and micro-seismic fluctuations. These sensor nodes collect physical index data and transmit it to a surface data processing center via an underground communication network. The control system then analyzes the underlying data, performs threshold comparisons, and renders it in 3D on terminal devices. When critical physical parameters exceed safety baselines, the system issues control commands to underground execution equipment to cut off power to the hazardous area and prevent the disaster from escalating. In this monitoring loop, the data acquisition frequency of the underlying sensors and the transmission scheduling mechanism of the communication backbone jointly determine the timeliness of the control commands, while the allocation of graphics rendering resources on the top-level visualization terminal directly affects the continuity and accuracy of the monitoring personnel's situational awareness at the disaster site.

[0003] Chinese invention patent application CN117689208A discloses a dynamic risk monitoring and intelligent classification early warning system for metal mines. The system includes components such as a data acquisition module, a distributed data storage center, a dynamic monitoring engine, and a risk monitoring panoramic view. The data acquisition module obtains real-time sampling data from various equipment at the mine site and uploads it to the distributed data storage center for transfer via a network transmission module. The dynamic monitoring engine retrieves data and rule units from the storage center and performs comparative analysis using a feature library model and basic high and low anomaly thresholds. When the monitored data falls within the abnormal alarm range, the system generates an early warning message, sends the alarm information to the user terminal using a notification broadcast module, and loads the early warning trigger distribution status of various scenes and equipment in the risk monitoring panoramic view using a 3D rendering engine.

[0004] In the initial, nonlinear, accelerated evolution phase of a mine disaster, the underlying sensor array generates high-frequency, bursty alarm data packets due to drastic fluctuations in monitored physical parameters. Existing mine safety systems typically employ a serial processing architecture, where data is first aggregated, then processed, and finally distributed for rendering and command issuance. Limited by the linear queuing mechanism of conventional network protocols, high-frequency alarm messages compete for bandwidth with regular monitoring data, resulting in severe transmission delays. This delay leads to physical control lag, occurring concurrently with the interruption of situational awareness for monitoring personnel caused by visualization rendering stutters in critical areas. Furthermore, there is a lack of direct hardware linkage between various safety subsystems based on underlying data flow. This serial delay disrupts the continuity of emergency command and decision-making, preventing surface monitoring personnel from timely assessing the underground disaster trend and from preemptively completing physical power-off isolation at the critical point of the disaster. Summary of the Invention

[0005] To address the control delays caused by message queuing and rendering stagnation, and to enable rapid physical power-off, this application provides a method and system for data communication and visualization of mine safety trend prediction.

[0006] Firstly, this application provides a method for data communication and visualization of mine safety trend prediction, employing the following technical solution: A method for data communication and visualization of mine safety trend prediction, applied to a mine safety monitoring system including a sensor array, calculation nodes, time-sensitive network switching equipment, visualization terminal, and explosion-proof execution equipment, comprising:

[0007] The measurement node acquires the real-time physical quantities of the sensor array using a preset first frequency, and calculates the trend approximation rate based on preset baseline values, limit values, anti-overflow constants and the first-order time derivative of the real-time physical quantities.

[0008] When the trend approximation rate is greater than the preset threshold corresponding to the real-time physical quantity type, the measurement node switches to a preset second frequency to acquire the real-time physical quantity and generate an alarm message. The alarm message encapsulates the target coordinates, peak time, network identifier, and redirection bit. Specifically, when the absolute value of the first-order time derivative is greater than the preset minimum effective slope, the peak time is obtained by dividing the difference between the limit value and the real-time physical quantity by the first-order time derivative; otherwise, the peak time is set to a preset upper limit value. When the calculated peak time is less than or equal to a preset lower limit value, the peak time is clamped to the preset lower limit value.

[0009] The time-sensitive network switching device identifies the network identifier, closes the transmission gate of the preset regular queue to truncate regular data, and establishes a dedicated channel to transmit the alarm message;

[0010] Based on the redirection position, the visualization terminal reduces the rendering accuracy of the 3D model of the area outside the target coordinates, and performs 3D rendering of the warning area according to the rendering refresh priority mapped by the peak time, so as to present the disaster evolution margin and the physical isolation state of power outage. The shorter the peak time, the higher the rendering refresh priority it is mapped to.

[0011] The explosion-proof actuator synchronously receives the alarm message through the dedicated channel and cuts off the power supply to the power supply zone or feeder circuit corresponding to the target coordinate.

[0012] Optionally, before the measurement node acquires the real-time physical quantities of the sensing array using the preset first frequency, the method further includes:

[0013] The visualization terminal preloads the full mesh constituting the three-dimensional model into the terminal's video memory and establishes a mapping relationship between the full mesh and spatial coordinates.

[0014] The calculation node reads the preset baseline value and the preset limit value from a storage medium, and loads the preset overflow prevention constant for calculation.

[0015] Optionally, the calculation of the trend approximation rate based on the preset baseline value, limit value, overflow prevention constant, and the first time derivative of the real-time physical quantity includes:

[0016] When the real-time physical quantity is an upper limit physical quantity that represents an increase in danger due to a rise in its value, the difference between the real-time physical quantity and the preset baseline value is calculated within the calculation node and used as a fluctuation term.

[0017] The difference between the preset limit value and the real-time physical quantity is added to the preset overflow prevention constant to obtain the margin value.

[0018] When the margin value is less than or equal to zero, the margin value is clamped to the preset anti-overflow constant;

[0019] When the direction of the first-order time derivative indicates that the real-time physical quantity is moving away from the preset limit value, the trend approximation rate is set to zero.

[0020] Otherwise, the trend approximation rate is obtained by multiplying the ratio of the fluctuation term to the clamped margin value by the first time derivative of the real-time physical quantity.

[0021] Optionally, within a preset period during which the measuring node switches from the preset first frequency to the preset second frequency, the method further includes:

[0022] The measurement node calculates the variance of the sampling sequence at the frequency intersection point, performs smoothing processing on the real-time physical quantity accordingly, and calculates the first-order time derivative based on the smoothed real-time physical quantity.

[0023] When the variance of the smoothed real-time physical quantity is higher than the preset variance threshold, or when the duration of the smoothing process exceeds the preset maximum suspension time, the calculation node performs calculations based on the highest historical effective derivative value instead of the first-order time derivative, or forcibly sets the peak time to a preset lower limit value.

[0024] Optionally, the generation of alarm messages by the measurement node includes:

[0025] The measuring node writes the network identifier into the link frame header of the alarm message to mark the highest transmission priority;

[0026] Write the redirection bit into the payload header of the alarm message, and write the peak time, the target coordinates, and the sampled sequence summary of the alarm trigger time into the payload of the alarm message.

[0027] Optionally, the step of having the time-sensitive network switching device identify the network identifier, close the transmission gate of a preset regular queue to truncate regular data, and establish a dedicated channel to transmit the alarm message includes:

[0028] The time-sensitive network switching device truncates the regular data based on the network identifier in the link frame header;

[0029] A frame preemption mechanism is adopted to suspend frame transmission in the preset regular queue in order to release bandwidth;

[0030] The target queue with the highest transmission priority is assigned exclusive bandwidth to send the alarm message.

[0031] Optionally, in the process of sending the alarm message with dedicated bandwidth, the method further includes:

[0032] When the buffer depth of the target queue reaches a preset congestion threshold due to uplink bandwidth bottleneck, the time-sensitive network switching device returns a suppression frame to the measurement node.

[0033] Based on the suppression frame, the measurement node reduces or discards the sampling sequence summary in the payload in the subsequently generated alarm message, while maintaining the integrity of the network identifier, the redirection bit, the target coordinates, and the peak time.

[0034] Optionally, the step of the visualization terminal reducing the rendering precision of the 3D model in the region outside the target coordinates based on the redirection bit, in order to release the graphics processor's computing power resources and improve the rendering capability of the warning area corresponding to the target coordinates, and performing 3D rendering of the warning area according to the rendering refresh priority mapped by the peak time, includes:

[0035] Based on the redirection bit, the geometric phase of the 3D model located outside the target coordinates is simplified to include distant mesh calculations, or complex shaders and post-processing effects of the 3D model in the outer region are disabled to reduce the shading calculation load on the graphics processor and to avoid triggering the unloading and reloading of the full mesh.

[0036] The graphics processor's rendering pipeline is focused onto the warning zone corresponding to the target coordinates;

[0037] Using the peak time as input, a preset rendering refresh priority mapping table is queried to obtain the corresponding rendering refresh priority, and the rendering pipeline is invoked to perform a high-priority thermal rendering task on the warning zone.

[0038] Optionally, during the execution of the high-priority thermal rendering task on the warning area, the method further includes:

[0039] When the terminal video memory occupancy rate exceeds a preset occupancy threshold within a preset period, causing the rendering thread to be suspended, a watchdog reset is triggered to reset the 3D rendering pipeline containing the rendering thread.

[0040] The vector interface is invoked to draw a two-dimensional topology map containing the target coordinates on the top layer of the screen.

[0041] Secondly, this application provides a mine safety trend prediction data communication and visualization system, which adopts the following technical solution: A mine safety trend prediction data communication and visualization system, the system comprising:

[0042] The physical quantity measurement module is used to acquire the real-time physical quantities of the sensor array using a preset first frequency, and to calculate the trend approximation rate based on preset baseline values, limit values, anti-overflow constants and the first-order time derivative of the real-time physical quantities.

[0043] An alarm generation module is used to switch to a preset second frequency to acquire the real-time physical quantity and generate an alarm message when the trend approximation rate is greater than a preset threshold corresponding to the real-time physical quantity type. The alarm message encapsulates the target coordinates, peak time, network identifier, and redirection bit. Specifically, when the absolute value of the first-order time derivative is greater than a preset minimum effective slope, the peak time is obtained by dividing the difference between the limit value and the real-time physical quantity by the first-order time derivative; otherwise, the peak time is set to a preset upper limit value. When the calculated peak time is less than or equal to a preset lower limit value, the peak time is clamped to the preset lower limit value.

[0044] The dedicated transmission module is used to identify the network identifier, close the transmission gate of the preset regular queue to truncate regular data, and establish a dedicated channel to transmit the alarm message;

[0045] The rendering execution module is used to reduce the rendering accuracy of the three-dimensional model of the region outside the target coordinates based on the redirection bit, and to perform three-dimensional rendering of the warning area according to the rendering refresh priority mapped by the peak time, so as to present the disaster evolution margin and the power outage physical isolation state. The shorter the peak time, the higher the rendering refresh priority it is mapped to.

[0046] The power supply interruption module is used to synchronously receive the alarm message through the exclusive channel and cut off the power supply to the power supply zone or feeder circuit corresponding to the target coordinate.

[0047] In summary, this application includes the following beneficial technical effects:

[0048] 1. By measuring the sampling frequency switching and alarm message generation triggered by the trend approximation rate of the node, and encapsulating the network identifier and redirection bit in the message, the time-sensitive network switching device can truncate the regular data and establish a dedicated transmission channel in real time according to the network identifier. The visualization terminal focuses the computing power on the warning zone for high-priority rendering in real time according to the redirection bit. The explosion-proof execution device synchronously parses the message and executes the power outage. In this way, the communication scheduling triggered by the alarm and the control of graphics rendering resources are coupled into a causal and collaborative link, avoiding the physical control and situational awareness delays caused by alarm message queuing and the dispersion of rendering computing power. This achieves rapid synchronous response of power outage command and warning screen in the critical stage of disaster.

[0049] 2. An overflow prevention constant is introduced into the trend approximation rate calculation, and the margin value is clamped. When the difference between the limit value and the real-time physical quantity approaches zero or becomes negative, the constant is forcibly used to replace it, preventing the approximation rate calculation failure caused by division by zero or numerical explosion. At the same time, combined with the direction determination of the first-order time derivative, the trend approximation rate when it is far from the limit value is directly set to zero, ensuring that high-frequency sampling and alarms are triggered only when the physical quantity is actually close to the danger threshold. This effectively reduces the probability of false alarms caused by calculation anomalies or trend misjudgment, and improves the accuracy and reliability of trend warning.

[0050] 3. The full mesh is preloaded to the terminal's video memory through the visualization terminal, and only the calculation of the peripheral distant mesh is simplified or complex shaders are disabled when an alarm is triggered without triggering mesh unloading and reloading. This avoids the bottleneck of video memory page swapping and bus transmission during the graphics processor's computing power redirection process. At the same time, when the video memory usage exceeds the standard and causes the rendering thread to hang, the watchdog is triggered to reset and the vector interface is called to draw a two-dimensional topology map containing the target coordinates as an emergency display. This ensures the immediacy of high-priority rendering switching in the warning area and the continuity of situational awareness under extreme resource shortages. Attached Figure Description

[0051] Figure 1 A flowchart illustrating a method for data communication and visualization in predicting mine safety trends, provided in an embodiment of this application;

[0052] Figure 2 This is a data structure diagram of the alarm message provided in the embodiments of this application. Detailed Implementation

[0053] The following combination Figures 1-2 This application will be described in further detail.

[0054] This application discloses a method for data communication and visualization in mine safety trend prediction. This method is applied to a mine safety monitoring system consisting of a sensor array, calculation nodes, a time-sensitive network switch, a visualization terminal, and explosion-proof execution equipment. The sensor array is deployed in the longwall face, return airway, transport roadway, goaf boundary, electromechanical chamber, and adjacent areas of the high-voltage power supply switch. The sensor array includes at least one of a gas concentration sensor, a micro-vibration detector, a roof strain sensor, a temperature sensor, a wind speed sensor, and a negative pressure sensor. The calculation nodes employ intrinsically safe edge computing equipment for mining, and are equipped with an analog-to-digital conversion unit, a clock synchronization unit, a trend gradient calculation unit, a frequency control unit, and a message encapsulation unit. The time-sensitive network switch uses an underground industrial switch supporting priority queuing and frame preemption mechanisms. The visualization terminal is equipped with a graphics processor and terminal video memory, and the explosion-proof execution equipment uses an intrinsically safe and explosion-proof power supply switch with a shunt trip mechanism or a vacuum circuit breaker mechanism.

[0055] During system initialization, the visualization terminal preloads the full mesh of the 3D model corresponding to mine roadways, working faces, equipment installation points, power supply zones, and emergency escape routes into its display memory. The full mesh consists of a vertex buffer, index buffer, material table, bounding box, and coordinate index table. The visualization terminal establishes a mapping relationship between the full mesh and spatial coordinates using the mine's unified coordinate system. Spatial coordinates include horizontal coordinates, vertical coordinates, and roadway mileage coordinates. For sensors underground that only have roadway mileage markers, the visualization terminal converts the roadway mileage markers into 3D coordinates using a pre-set roadway centerline table. For sensors with inertial navigation positioning or ultra-wideband positioning data, the visualization terminal directly uses the positioning coordinates as the 3D model mapping coordinates. These preloaded results serve as the input graphics resources for subsequent rendering.

[0056] During system initialization, the measurement nodes read from a storage medium the fixed baseline value, limit value, anti-overflow constant, multiple preset thresholds, preset minimum effective slope, first frequency, second frequency, preset variance threshold, preset longest hang time, preset congestion threshold, and preset occupancy threshold. The baseline value represents the safe operating reference value for the corresponding physical quantity, while the limit value represents the critical value that triggers hazard control for the corresponding physical quantity. For example, with gas concentration, the baseline value can be the historical safe operating average or the allowable value for mine ventilation design, and the limit value can be the concentration controlled during power outages. With microseismic energy, the baseline value can be the background energy average during stable mining, and the limit value can be the energy threshold corresponding to the roof impact hazard level. With roof strain, the baseline value can be the strain average during the stable support stage, and the limit value can be the strain threshold corresponding to roof delamination or support instability. The anti-overflow constant is a very small positive number greater than zero, used to prevent division by zero or numerical explosion when the difference between the limit value and the real-time physical quantity approaches zero or becomes negative. When using a standard measurement range of 0 to 100, the preferred value range for the overflow prevention constant is 0.1% to 1 / 100,000; when using a normalized measurement range of 0 to 1, the preferred value range for the overflow prevention constant is 1 / 100,000 to 1 / 1,000,000. The preset minimum effective slope is set based on the physical quantity characteristics and sensor noise level to prevent the peak time from becoming infinitely long when the derivative approaches zero. Multiple preset thresholds are independently set for different physical quantity types to determine whether their respective trend approximation rates meet the conditions for triggering an alarm.

[0057] The measurement node acquires real-time physical quantities from the sensor array using a preset first frequency. This first frequency is used for routine monitoring and is preferably from once per second to ten times per second. The sensor array converts gas concentration, microseismic velocity, microseismic energy, roof strain, temperature, or wind speed into electrical or digital signals. The measurement node receives these real-time physical quantities via an analog-to-digital converter or a digital bus interface and writes them into a sliding time window using a timestamp provided by a clock synchronization unit. The sliding time window preferably covers 500 milliseconds to 10 seconds, with the window length determined based on the rate of change of the physical quantities. For gas concentration and temperature, the window length is preferably 3 to 10 seconds; for microseismic events and roof strain, the window length is preferably 500 milliseconds to 3 seconds. The real-time physical quantities within the sliding time window serve as input to the trend gradient calculation unit.

[0058] The measurement node calculates the trend approximation rate based on preset baseline values, limit values, overflow prevention constants, and the first-order time derivatives of real-time physical quantities. This method differentiates between different types of physical quantities. For upper limit-type physical quantities such as gas concentration, microseismic energy, roof strain, or temperature, an increase in value indicates increased danger. The calculation process for the trend approximation rate is as follows: The measurement node calculates the difference between the real-time physical quantity and the baseline value within the trend gradient calculation unit to obtain the fluctuation term; the measurement node calculates the difference between the limit value and the real-time physical quantity, and adds this difference to the overflow prevention constant to obtain the margin value. When this margin value is less than or equal to zero, the measurement node forcibly clamps this margin value to the overflow prevention constant to ensure the validity of subsequent calculations. Subsequently, the measurement node judges the trend direction of the real-time physical quantity. When the direction of the first-order time derivative indicates that the real-time physical quantity is moving away from the limit value, it means that the environment is approaching safety, and the measurement node directly sets the trend approximation rate to zero to avoid generating false alarms. Otherwise, when a physical quantity is approaching its limit, the measurement node calculates the ratio of the fluctuation term to the clamped margin value, and multiplies this ratio by the first-order time derivative of the real-time physical quantity to obtain an effective trend approximation rate. For the aforementioned upper limit type physical quantities, the trend approximation rate... The specific calculation formula is as follows:

[0059]

[0060] In the formula, For real-time physical quantities, Baseline value, For the limit value, As an overflow prevention constant, This refers to the first-order time derivative of the real-time physical quantity. Through the underlying combination of the aforementioned anti-overflow constant clamping mechanism and the first-order time derivative direction determination logic, the system can effectively filter instantaneous high-frequency environmental noise and non-trend mutations during the front-end calculation stage, avoiding false power outages caused by false alarms, thereby ensuring the stable operation of related safety circuits such as mine ventilation and drainage from the source. For lower limit physical quantities such as wind speed and negative pressure, a decrease in value indicates an increase in danger, and the margin value is formed by adding the difference between the real-time physical quantity and the limit value to the anti-overflow constant. When this margin value is less than or equal to zero, it is also clamped to the anti-overflow constant. The fluctuation term is determined by the difference between the baseline value and the real-time physical quantity. When determining the direction, if the direction of the first-order time derivative indicates that the value is rising and moving away from the danger lower limit, the trend approximation rate is set to zero; otherwise, the ratio of the fluctuation term to the clamped margin value is multiplied by the absolute value of the first-order time derivative of the real-time physical quantity to obtain the positive trend approximation rate. The first-order time derivative of a real-time physical quantity can be calculated using the difference between adjacent sampling points or the least-squares slope within a sliding window. To suppress downhole electromagnetic interference and sensor quantization noise, the measurement node can perform median filtering, amplitude limiting filtering, or Kalman filtering on the real-time physical quantity before differentiation.

[0061] When the trend approximation rate exceeds the preset threshold corresponding to the real-time physical quantity type, the frequency control unit of the measurement node switches the sampling frequency from the first frequency to the second frequency. The second frequency is used for high-frequency monitoring during the accelerated phase of the disaster trend, preferably from fifty to two hundred times per second. The preset thresholds for each type of physical quantity are obtained by back-calculation from their respective historical disaster samples, field calibration data, or safety procedure thresholds, and are independent of each other, without cross-dimensional comparisons or fusion. For example, the preset threshold for gas concentration can be determined by the historical upward slope when the safe concentration approaches the power outage concentration; the preset threshold for microseismic energy can be determined by the energy growth rate of precursory rockburst events. Each preset threshold can also be set as a graded threshold. When the trend approximation rate exceeds the first-level threshold, only the sampling frequency is increased; when the trend approximation rate exceeds the second-level threshold, the sampling frequency is increased and an alarm message is triggered; when the trend approximation rate exceeds the third-level threshold, the sampling frequency is increased, an alarm message is triggered, and a forced power outage is triggered simultaneously.

[0062] Within a preset period during the frequency transition from the first to the second frequency, the measurement node calculates the variance of the sampled sequence at the frequency transition point and performs smoothing on the real-time physical quantity accordingly. Based on the smoothed real-time physical quantity, it calculates the first-order time derivative. The smoothing process can employ a Kalman filter model, an exponentially weighted model, or a sliding window model. When the variance of the smoothed real-time physical quantity exceeds a preset variance threshold, it indicates that the data sequence has exhibited drastic fluctuations and entered a catastrophic accelerated evolution phase. Alternatively, when the smoothing process lasts longer than a preset maximum suspension time, to prevent direct differentiation of the original data filled with high-frequency noise from causing a nonlinear numerical explosion that could lead to system deadlock, the measurement node calculates the first-order time derivative based on the highest historical effective derivative value, or forcibly sets the peak time to a preset lower limit to ensure the trend determination remains valid.

[0063] When the trend approximation rate of any physical quantity exceeds its corresponding preset threshold, the measurement node generates an alarm message. The alarm message encapsulates the target coordinates, peak time, network identifier, and redirection bit. The target coordinates are determined by the sensor installation coordinates that triggered the limit exceedance, the grid coordinates to which the sensor belongs, or the power supply zone coordinates to which the sensor belongs. The peak time represents the estimated time required for the real-time physical quantity to approach the limit value according to the current trend.

[0064] During the catastrophe evolution, the trajectory of real-time physical quantities changes over time, crossing a preset baseline value and accelerating towards a preset limit value. The first-order time derivative objectively characterizes the instantaneous rate of change of the physical quantity at the current sampling moment. The sampling action of the measurement node corresponds to a preset first frequency (i.e., normal sparse sampling) before the frequency transition, and a more densely packed preset second frequency after the frequency transition. When the absolute value of the first-order time derivative is greater than a preset minimum effective slope, the peak time essentially represents the remaining time obtained by linear extrapolation prediction based on the current instantaneous rate of change, specifically calculated by dividing the difference between the limit value and the real-time physical quantity by the first-order time derivative; otherwise, the peak time is set to a preset upper limit value.

[0065] When the calculated peak time is less than or equal to a preset lower limit, the peak time is clamped to the preset lower limit to prevent over-response. The peak time can also be obtained by extrapolating a quadratic fitted curve within a sliding window, or by using a lightweight regression model trained on historical disaster samples.

[0066] like Figure 2 As shown, the alarm message adopts a unified encapsulation format for the link frame header and application payload, and is divided into three blocks arranged independently in sequence: the link frame header, the payload header, and the payload. In specific communication engineering implementations, the measuring node nests the network identifier as an independent field within the link frame header of the alarm message to mark the highest transmission priority. For example, this network identifier can map to and occupy the 3-bit priority bit field in the virtual LAN label conforming to the IEEE 802.1Q standard. After the network identifier is written, the system locks the highest priority queue. The measuring node writes the redirection bit separately into the payload header of the alarm message, for example, occupying one byte at the beginning of the payload header. It should be noted that the redirection bit defined here does not refer to the address jump identifier at the conventional computer network routing layer, but specifically refers to a hardware control status bit defined in the alarm message to trigger the visualization terminal to forcibly transfer the graphics processor's rendering pipeline computing power from the global conventional area and focus it on the target warning area. Subsequently, the measuring node concatenates the sampled sequence summaries of the target coordinates, peak time, and alarm trigger time in a compact, side-by-side order from left to right, and writes them as horizontal parallel fields into the end payload of the alarm message. Each field occupies a preset fixed byte length. The payload header may also include protocol version, message type, coordinate type, peak time length, checksum field, and reserved fields.

[0067] The redirection bit is used to instruct the visualization terminal to perform graphics computing power redirection. The payload may also carry sensor number, real-time physical quantity, trend approximation rate, power supply zone number, and timestamp. To ensure reliable resolution, alarm messages can use a cyclic redundancy check field, a sequence number field, and a duplicate transmission flag field.

[0068] Time-Sensitive Network Switching (TSNS) devices receive alarm messages and identify network identifiers. The ingress port of the TNS parses the network identifier within the link frame header and maps the alarm message with the highest transmission priority to the target queue. The TNS truncates regular data based on the network identifier in the link frame header. Regular data includes periodic inspection data, ordinary video streams, device status data, and low-priority historical log data. Truncation is accomplished through a frame preemption mechanism, whereby the TNS suspends frame transmission in the low-priority regular queue, allowing the target queue mapped to the highest transmission priority to send alarm messages with dedicated bandwidth. This dedicated channel prioritizes alarm messages as the sole transmission target, ensuring that critical alarms and control commands can be sent with minimal latency, avoiding bandwidth competition between alarm messages and regular data in the linear queue, thus providing communication assurance for the rapid power-off of explosion-proof equipment.

[0069] During the process of sending alarm messages with dedicated bandwidth, the time-sensitive network switch monitors the buffer depth of the target queue. When the buffer depth of the target queue reaches a preset congestion threshold due to uplink bandwidth bottlenecks, the time-sensitive network switch returns a suppression frame to the measurement node. The preset congestion threshold is preferably 70% to 90% of the target queue's buffer capacity. The suppression frame carries a congestion flag, queue number, buffer depth, and allowed packet length. Based on the suppression frame, the measurement node reduces or completely discards non-critical sample sequence summary data in the payload when generating new alarm messages to reduce packet size; the downgraded packet must maintain the integrity of the network identifier, redirection bit, target coordinates, and peak time to ensure that core control information is not lost.

[0070] The visualization terminal receives alarm messages and parses the redirection bit, target coordinates, and peak time. Based on the redirection bit, the visualization terminal reduces the rendering precision of the 3D model in the area outside the target coordinates to free up graphics processor (GPU) computing resources. These freed resources are then used to increase the rendering refresh priority of the warning zone corresponding to the target coordinates, ensuring smooth, lag-free visuals in critical areas. The specific method of reducing rendering precision is as follows: based on the redirection bit, the visualization terminal simplifies the distant mesh calculation during the geometric stage of the 3D model in the area outside the target coordinates, or disables complex shaders and post-processing effects for the 3D model in the outer area. This reduces the vertex shading and texture sampling calculations required by the GPU, freeing up GPU computing resources. This process does not trigger the unloading and reloading of the entire mesh, preventing bus congestion. The area outside the target coordinates can be determined by the warning radius centered on the target coordinates. The warning radius can be determined based on the type of physical quantity, tunnel topology, and power supply zone. For gas concentration, the warning radius preferably covers the ventilation zone where the target coordinates are located; for micro-seismic events and roof strain, the warning radius preferably covers the support unit where the target coordinates are located and adjacent support units.

[0071] The visualization terminal focuses the released graphics processing power on the warning zone corresponding to the target coordinates. The warning zone is jointly determined by the target coordinates, warning radius, power supply zone, and roadway topology. The visualization terminal uses the peak time as input and queries a preset rendering refresh priority mapping table to obtain the corresponding rendering refresh priority. This mapping ensures that the shorter the peak time and the higher the risk, the higher the priority of the system's rendering task scheduling for that warning zone, thus guaranteeing timely human-computer interaction feedback. The rendering pipeline allocates computing power according to this priority, performing thermal rendering on the warning zone to present the disaster evolution margin and the physical isolation state of the power outage. Thermal rendering includes applying color gradients, pulse brightness, risk isosurfaces, hazard diffusion direction arrows, and isolation zone markings to the grid near the target coordinates. The color mapping for thermal rendering can use a continuous color band from cool to warm colors, or a single-color highlight overlaid with transparent pulses. The visualization terminal can also map the trend approximation rate to thermal intensity, the peak time to the dynamic update rate of the warning markers, and the target coordinates to three-dimensional anchor points.

[0072] During the high-priority thermal rendering task of the warning zone, the visualization terminal monitors the terminal's video memory usage and rendering thread status. When the terminal's video memory usage exceeds a preset threshold within a preset period, causing the rendering thread to suspend, the visualization terminal triggers a watchdog reset to reset the 3D rendering pipeline containing the rendering thread and switches to the 2D emergency display channel. The preset period is preferably three to ten display refresh cycles, and the preset usage threshold is preferably 90% to 98% of the video memory capacity. The watchdog reset can clear the 3D command queue, release abnormal texture resources, and disable complex shaders. The visualization terminal calls the vector interface to draw a 2D topology map containing target coordinates on the top layer of the screen. The 2D topology map includes lane segments, sensor identifiers, power supply zone boundaries, target coordinate markers, warning margin markers, and power outage status markers. As an equivalent alternative, the visualization terminal can call the web canvas interface, operating system graphics device interface, or embedded frame buffer interface to draw the 2D topology map; the visualization terminal can also call a pre-generated raster base map and overlay the target coordinates on the raster base map when the 3D model is unavailable.

[0073] Time-sensitive network switching equipment synchronously outputs alarm messages to the explosion-proof execution equipment via a dedicated channel. The explosion-proof execution equipment parses the target coordinates or power supply zone number in the alarm message and maps the target coordinates to the corresponding feeder switch, branch switch, or area power-off circuit. Based on the alarm message, the explosion-proof execution equipment triggers the shunt trip mechanism, vacuum circuit breaker mechanism, or solid-state power-off module to cut off the power supply to the power supply zone or feeder circuit corresponding to the target coordinates. The explosion-proof execution equipment detects the power-off status through a current transformer, voltage sampling circuit, or switch auxiliary contact and writes the power-off status receipt into a receipt message. The receipt message can be returned to the calculation node and visualization terminal along the dedicated channel for displaying the power-off completion status on the visualization terminal. For mines with multi-level power supply structures, the explosion-proof execution equipment prioritizes cutting off the lower-level branch corresponding to the target coordinates; when the lower-level branch feedback is abnormal or fails to operate, the explosion-proof execution equipment cuts off the upper-level interconnection switch or main feeder switch. Both the physical power failure output and the 3D rendering output use the same alarm message as input. The dedicated TSN channel ensures extremely low latency transmission of power failure commands, solving the problem of physical control lag. Meanwhile, the 3D rendering output ensures the smoothness of the monitoring screen and accurately maps the power failure state through computing power redirection, solving the problem of situational awareness interruption. The two technologies work in parallel, forming a causal closed loop of data flow from sensing calculation, message encapsulation, dedicated transmission to computing power redirection and physical power failure.

[0074] In one optional implementation, the sensor array includes a laser spectral gas probe and a piezoelectric microseismic detector. The laser spectral gas probe outputs gas concentration, and the piezoelectric microseismic detector outputs microseismic velocity or microseismic energy. The measurement node calculates the trend approximation rate for both gas concentration and microseismic energy, and generates an alarm message when either trend approximation rate exceeds its corresponding preset threshold. The target coordinates preferentially use the sensor coordinates that triggered the alarm, and the peak time preferentially uses the calculated value corresponding to the alarm trigger. This method is suitable for fully mechanized mining faces where gas anomalies are coupled with roof impact risks.

[0075] In another optional implementation, the sensing array includes fiber Bragg grating strain sensors. These sensors are deployed along the roof anchor cables, surrounding rock of the tunnel, or support beams. The measurement nodes use the strain values ​​demodulated by the fiber Bragg gratings as real-time physical quantities. The baseline value is the average strain under stable support conditions, and the limiting value is the strain threshold corresponding to roof delamination or support instability. The measurement nodes trigger alarm messages based on the trend approximation rate calculation method corresponding to the strain values. This method is suitable for scenarios involving roof deformation, tunnel surrounding rock instability, and abnormal support structures.

[0076] In another alternative implementation, the sensor array includes an infrared thermal imaging device. The infrared thermal imaging device outputs the surface temperature or temperature rise rate of the coal body. The calculation node uses the temperature of the target area as a real-time physical quantity, with the baseline value being the average temperature under normal ventilation and mining conditions, and the extreme value being the coal spontaneous combustion warning temperature or the equipment overheating power-off temperature. The calculation node identifies the stage of continuously accelerating temperature rise through the trend approximation rate and encapsulates the location of the thermal anomaly as the target coordinates into an alarm message. This method is suitable for scenarios involving coal spontaneous combustion, belt friction overheating, and electromechanical equipment overheating.

[0077] In mines lacking full Time-Sensitive Networking (TSN) capabilities, TSN switches can be replaced with switches supporting differential services, Quality of Service (QoS) queuing, or Industrial Ethernet priority queuing. The measurement node writes the network identifier into a differential service field or a custom priority field. The switch uses this field to map alarm messages to the highest priority queue and suspends or delays dequeueing from ordinary queues. This alternative method still uses the network identifier as input for regular data truncation and alarm message priority transmission. For wireless backhaul links, the network identifier can be mapped to the highest access class in wireless resource scheduling; redirection bits, peak arrival time, and target coordinates are still transmitted in the payload.

[0078] Combination Figure 1 As shown, the measurement node continuously calculates the trend approximation rate and performs logical judgments. If it is not greater than the corresponding preset threshold, it returns to the loop for measurement. When it is greater than the preset threshold, the trend approximation rate output by the measurement node directly triggers high-frequency sampling and generates an alarm message containing the encapsulated target coordinates and peak time action through the above method. The time-sensitive network switching device identifies the network identifier and closes the preset regular queue transmission gate, thereby directly driving the establishment of an exclusive channel to transmit the alarm message. Subsequently, the transmission link is divided into two parallel branches: in the left branch, the redirection bit, peak time, and target coordinates carried in the alarm message directly drive the visualization terminal to reduce the rendering accuracy of the area outside the target coordinates and focus on the rendering of the warning area according to the peak time mapping priority; in the right branch, the target coordinates or power supply zone number carried in the alarm message directly drive the explosion-proof execution device to cut off the power supply. Finally, the two control branches converge to uniformly present the disaster evolution margin and the physical isolation status of the power outage. Each step has a clear physical execution subject, and the upstream output is used as the downstream input condition, which can realize rapid physical power outage response and smooth situational awareness screen guarantee in the critical stage of disaster in parallel.

[0079] This application also discloses a mine safety trend prediction data communication and visualization system. This system is applicable to executing the aforementioned mine safety trend prediction data communication and visualization method. The system includes a virtual functional module architecture and a physical hardware architecture. The virtual functional module architecture represents the functional coupling relationship between data input, data processing, and data output. The physical hardware architecture represents the physical bearing relationship between the processor, memory, communication interface, graphics processor, switching chip, and explosion-proof actuator. The functions of the virtual functional modules are implemented by the processor, graphics processor, switching chip, or programmable logic device in the physical hardware architecture calling program instructions, firmware entries, and configuration parameters in the memory, or are solidified in the physical hardware architecture as software, firmware, hardware logic, or a combination thereof.

[0080] In the virtual functional module architecture, the system includes a physical quantity measurement module, an alarm generation module, a dedicated transmission module, a rendering execution module, a power supply interruption module, a resource residency module, a frequency conversion smoothing module, a congestion degradation module, and an emergency topology module.

[0081] The resource-resident module is configured to: receive the full mesh corresponding to mine roadways, equipment, power supply zones, and sensor installation points as 3D model input; receive sensor coordinates, roadway mileage coordinates, power supply zone coordinates, and mine unified coordinate system parameters as spatial coordinate input; and receive baseline values, limit values, overflow prevention constants, multiple preset thresholds, preset minimum effective slope, first frequency, second frequency, preset variance threshold, preset longest hang time, preset congestion threshold, and preset occupancy threshold as firmware parameter input. The resource-resident module outputs the mapping between the full mesh and spatial coordinates to the video memory, and outputs calculation parameters such as baseline values, limit values, overflow prevention constants, and multiple preset thresholds. The mapping in the video memory serves as the mesh resource input for the rendering execution module, and the calculation parameters serve as the parameter input for the physical quantity calculation module.

[0082] The physical quantity measurement module is configured to: receive real-time physical quantities from the sensor array as sensor data input; receive baseline values, limit values, overflow prevention constants, and multiple preset thresholds output by the resource-resident module as measurement parameter inputs; and utilize a time window to cache and store the real-time physical quantity sequence with timestamps. The physical quantity measurement module includes functions such as fluctuation term calculation, margin value calculation, derivative calculation, clamping, direction determination, and approximation rate synthesis. The fluctuation term calculation outputs the difference between the real-time physical quantity and the baseline value. The margin value calculation differentiates processing based on the physical quantity type: for upper limit physical quantities, it outputs the sum of the difference between the limit value and the real-time physical quantity and the overflow prevention constant; for lower limit physical quantities, it outputs the sum of the difference between the real-time physical quantity and the limit value and the overflow prevention constant. The clamping function forces the margin value to be set to the overflow prevention constant when the margin value is less than or equal to zero. The derivative calculation outputs the first-order time derivative of the real-time physical quantity. The direction determination function determines the trend direction based on the sign of the first-order time derivative and the physical quantity type. The output of the approximation rate synthesis is as follows: when the direction is determined to be far from the limit value, it is set to zero; when the direction is determined to be close to the limit value, for upper limit physical quantities, the ratio of the fluctuation term to the clamped margin value is multiplied by the first-order time derivative; for lower limit physical quantities, the ratio of the fluctuation term to the clamped margin value is multiplied by the absolute value of the first-order time derivative. The trend approximation rate output by the physical quantity measurement module serves as the trigger input for the alarm generation module. Different types of physical quantities generate their own trend approximation rates independently and compare them with their corresponding preset thresholds.

[0083] The frequency conversion smoothing module is configured to receive the frequency switching status from the alarm generation module and the frequency transition point samples from the time window buffer of the physical quantity measurement module. The module calculates the variance of the transition point samples and times the smoothing processing duration. The smoothed real-time physical quantity output is used to calculate the first-order time derivative and sent to the derivative calculation stage of the physical quantity measurement module. When the variance exceeds a preset variance threshold or the smoothing processing duration exceeds a preset maximum suspension time, the derivative calculation is performed directly based on the highest historical effective derivative value instead of the first-order time derivative. The frequency conversion smoothing module may include a Kalman filter unit, an exponentially weighted variance unit, or a sliding window variance unit.

[0084] The alarm generation module is configured to: receive the trend approximation rate output by the physical quantity measurement module as a trigger input; output the sampling control signal corresponding to the second frequency; receive sensor coordinates, hazard center coordinates, or power supply zone coordinates; and receive the peak time determined by real-time physical quantities, first-order time derivatives, preset minimum effective slope, and limit values. The alarm generation module outputs alarm messages, network identifiers, and redirection bits. The alarm generation module includes functions such as frequency control, peak time estimation, link frame header encapsulation, and payload encapsulation. Link frame header encapsulation outputs the link frame header with the highest transmission priority. Payload encapsulation outputs a payload containing redirection bits, target coordinates, peak time, and a sampling sequence summary. The alarm message serves as the message input for the dedicated transmission module.

[0085] The dedicated transmission module is configured to: receive alarm messages output by the alarm generation module; parse the highest transmission priority within the link frame header; control frame preemption to suspend frame transmission in a preset regular queue; output alarm messages with dedicated bandwidth to the target queue; and output suppression frames when the target queue buffer depth reaches a preset congestion threshold due to aggregation bandwidth limitations. The dedicated transmission module has dual-path distribution capabilities, enabling it to output alarm messages to both the visualization terminal and the explosion-proof execution device. The dedicated transmission module includes functions such as priority mapping, frame preemption execution, target queue scheduling, and buffer depth monitoring. The alarm messages output by the target queue serve as input to the rendering execution module and the power interruption module.

[0086] The congestion reduction module is configured to receive suppression frames from the dedicated transmission module and raw alarm messages from the alarm generation module. While maintaining the integrity of necessary fields such as network identifier, redirection bit, peak arrival time, and target coordinates, it outputs a reduced or discarded sample sequence summary as a degraded message. This degraded message serves as the message input for the dedicated transmission module.

[0087] The rendering execution module is configured to: receive alarm messages output by the exclusive transmission module; parse the redirection bits in the payload header; extract the target coordinates and peak time from the payload; and receive the mapping association between the full mesh and spatial coordinates output by the resource resident module. The rendering execution module outputs a computing power focusing control command to concentrate the graphics processor's computing power on the warning zone corresponding to the target coordinates, and outputs a 3D thermal image of the warning zone to reflect the evolution of the disaster and the power outage status. The rendering execution module includes functions such as region determination, rendering calculation simplification, computing power redirection, rendering refresh priority mapping, and thermal shading. The rendering refresh priority mapping uses the peak time as an index to look up a preset mapping table to obtain the corresponding rendering refresh priority, and schedules the rendering pipeline to execute high-priority tasks accordingly. The thermal rendering image is output to the display device. The rendering calculation simplification function frees up computing power by simplifying the distant mesh calculation or disabling peripheral effects during the geometry stage.

[0088] The emergency topology module is configured to receive terminal memory usage and rendering thread suspension status. When trigger conditions are met, its watchdog control outputs a reset control signal to stop or reset the 3D rendering pipeline. The emergency topology module stores 2D topology resources such as tunnel segments, power supply zones, sensor locations, and target coordinate markers, and can draw a 2D topology map containing target coordinates on the top layer of the screen. The top-layer output is connected to the display device. The emergency topology module may include functions such as vector drawing, raster overlay, and power failure status annotation.

[0089] The power supply interruption module is configured to: receive alarm messages output by the exclusive transmission module; parse and extract target coordinates; map the target coordinates to a feeder switch, branch switch, or area power outage circuit; output a power outage control signal to cut off the power supply corresponding to the target coordinates; and output receipt information such as power outage status, switch status, current status, and voltage status. The power outage receipt can be returned to the exclusive transmission module, rendering execution module, or physical quantity measurement module. The power supply interruption module may include functions such as shunt trip control, vacuum circuit breaker control, solid-state power outage control, and phase current feedback.

[0090] In the virtual functional module architecture, the trend approximation rate of the physical quantity measurement module serves as the trigger input of the alarm generation module, the alarm message of the alarm generation module serves as the message input of the exclusive transmission module, the dual-path distribution output of the exclusive transmission module serves as the alarm input of the rendering execution module and the power interruption module, the thermal rendering output of the rendering execution module serves as the graphic input of the display device, and the power failure receipt of the power interruption module serves as the visual status display input.

[0091] In the physical hardware architecture, the system includes a sensor array, an intrinsically safe edge prediction node for mining, a time-sensitive network switch, a twin visualization control terminal, an explosion-proof physical execution unit, and a display device.

[0092] The sensor array includes at least one of the following: a gas concentration sensor, a micro-vibration detector, a fiber optic strain sensor, a temperature sensor, a wind speed sensor, and a negative pressure sensor. The sensor array has a sensing probe, a signal conditioning circuit, an analog-to-digital converter, a digital communication interface, and an explosion-proof housing. The digital communication interface of the sensor array can be an intrinsically safe Ethernet interface, a controller area network (CLAN) interface, a serial bus interface, a fiber optic interface, or a wireless intrinsically safe communication interface. The output of the sensor array is electrically or communicatively connected to the sensing input of the intrinsically safe edge prediction node for mining applications.

[0093] The intrinsically safe edge prediction node for mining includes a first processor, a first memory, a clock synchronization circuit, a sensor interface, a trend gradient arithmetic logic unit, a frequency control circuit, a message encapsulation circuit, and a first network interface. The first processor can be a low-power microprocessor, a digital signal processor, a field-programmable gate array (FPGA), or a system-on-a-chip (SoC). The first memory stores baseline values, limit values, anti-overflow constants, multiple preset thresholds, a preset minimum effective slope, a first frequency, a second frequency, a preset variance threshold, a preset maximum suspend time, and the message encapsulation program. The trend gradient arithmetic logic unit includes subtractors, adders, dividers, multipliers, registers, comparators, clamping logic, and fixed-point or floating-point anti-overflow processing circuits, and is configured with logic to select the appropriate fluctuation term based on the physical quantity type, perform margin value calculations, and formulate the approximation rate synthesis formula. The frequency control circuit is connected to the sensor interface and has a first frequency configuration register and a second frequency configuration register. The message encapsulation circuit has link frame header writing logic and payload writing logic. The first network interface is connected to the ingress port of a time-sensitive network switch. The functions of the physical quantity measurement module, alarm generation module, frequency conversion smoothing module, and congestion degradation module are executed by the first processor calling program instructions in the first memory, or are solidified in hardware logic by the trend gradient arithmetic logic unit, frequency control circuit, and message encapsulation circuit.

[0094] The time-sensitive network switch includes a switching chip, queue buffer, frame preemption control circuit, multiple physical ports, a time synchronization unit, and a priority scheduling circuit. The switching chip supports Ethernet priority mapping, time-aware shaping, frame preemption, or quality of service scheduling. The queue buffer includes multiple traffic queues, at least one of which is the highest priority target queue, while the remaining queues are regular queues. The frame preemption control circuit is connected to the transmitting end of each traffic queue to pause the transmission of low-priority frames. The time synchronization unit can use a precise time protocol clock, a crystal oscillator clock, or a unified mine time synchronization module. The input ports of the switching chip are connected to intrinsically safe edge prediction nodes for mining, and the output ports are connected to a twin visualization control terminal and an explosion-proof physical execution unit, respectively. The function of the dedicated transmission module is executed by the switching chip calling the queue buffer and frame preemption control circuit, or it can be implemented by the priority scheduling circuit and frame preemption control circuit in a hardware logic manner.

[0095] The twin visualization control terminal includes a second processor, a second memory, a graphics processor, terminal video memory, a display interface, a second network interface, and a watchdog circuit. The second processor can be an industrial PC processor, an embedded processor, or a server processor. The second memory stores the 3D mine model, a spatial coordinate mapping table, the rendering engine program, a rendering refresh priority mapping table, 2D topology resources, and an emergency display program. The graphics processor includes shader cores, a command processor, texture units, rasterization units, and rendering output units. The terminal video memory stores the full mesh, material tables, texture resources, vertex buffers, index buffers, and scene graph nodes. The display interface connects to a display device. The second network interface connects to a time-sensitive network switch. The watchdog circuit connects to the second processor and the graphics processor. The functions of the resource resident module, rendering execution module, and emergency topology module are executed by the second processor calling program instructions in the second memory, and implemented by the graphics processor calling graphics resources in the terminal video memory; among them, the distant mesh simplification, computing power redirection, rendering refresh priority mapping, and thermal rendering can be executed by the command processor, shader core, and rendering pipeline of the graphics processor, and the two-dimensional topology drawing can be executed by the second processor calling the operating system graphics interface, vector interface, or frame buffer interface.

[0096] The explosion-proof physical execution unit includes a third processor, a third memory, a third network interface, a coordinate partition mapping table, a trip drive circuit, a circuit breaker mechanism, a voltage sampling circuit, a current transformer, and an auxiliary contact feedback circuit. The third processor can be a mining controller, a programmable logic controller, or an embedded microcontroller. The third memory stores the mapping table between the target coordinates and the power supply partition, the power outage control program, the switch failure judgment parameters, and the acknowledgment message format. The third network interface connects to a time-sensitive network switch. The trip drive circuit connects to the circuit breaker mechanism. The circuit breaker mechanism can be a shunt trip mechanism, a vacuum circuit breaker mechanism, or a solid-state switch assembly. The voltage sampling circuit and current transformer are used to detect the voltage and current status after power outage. The auxiliary contact feedback circuit is used to detect the switch's open / closed state. The power supply interruption module's function is executed by the third processor calling program instructions from the third memory, and the physical power outage and status feedback are completed by the trip drive circuit, circuit breaker mechanism, voltage sampling circuit, current transformer, and auxiliary contact feedback circuit.

[0097] The display equipment includes at least one of the following: a mine display screen, a dispatch center display, an intrinsically safe underground display terminal, or a mobile inspection terminal. The display equipment has a 3D early warning screen input interface and a 2D topology screen input interface. The 3D early warning screen input interface receives the thermal rendering screen from the twin visualization control terminal. The 2D topology screen input interface receives the 2D topology map output by the emergency topology module. The display equipment can display target coordinates, peak time, trend approximation rate, power outage status, power supply zone, and sensor number.

[0098] In the physical hardware architecture, the real-time physical quantity output of the sensor array is connected to the sensor input of the intrinsically safe edge prediction node for mining. The first network interface of the intrinsically safe edge prediction node is connected to the input port of the time-sensitive network switch. The output port of the time-sensitive network switch is connected to the second network interface of the twin visualization control terminal and the third network interface of the explosion-proof physical execution unit, respectively. The display interface of the twin visualization control terminal is connected to the display device. The power-off receipt output of the explosion-proof physical execution unit is connected back to the time-sensitive network switch via the third network interface. The hardware connection relationship ensures that the real-time physical quantity, trend approximation rate, alarm message, rendering control field, and power-off control field maintain a closed loop in the physical link.

[0099] In a preferred physical configuration, the first processor of the intrinsically safe edge prediction node for mining uses a microprocessor with a floating-point unit, the trend gradient arithmetic logic unit uses parallel computing logic in a field-programmable gate array (FPGA), and the first memory uses a combination of non-volatile memory and random access memory (RAM). The time-sensitive network switch has eight traffic queues; the highest priority target queue maps to the highest priority network identifier in the link frame header, while the regular queues carry video inspection data, environmental inspection data, and equipment log data. The twin visualization control terminal is equipped with an independent graphics processor and dedicated terminal video memory, and the 3D model resides in vertex buffers and index buffers. The explosion-proof physical execution unit is equipped with a shunt trip mechanism and a current inductor feedback loop. This configuration is suitable for deep fully mechanized mining faces with high requirements for disaster response time.

[0100] In a compatible entity configuration, the time-sensitive network switch can be replaced with an industrial Ethernet switch that supports differential service or quality of service scheduling, and the frame preemption control circuit can be replaced with a strict priority queue scheduling circuit. The packet encapsulation circuit of the intrinsically safe edge prediction node for mining writes the network identifier into the differential service field or a custom priority field. This configuration is suitable for upgrading existing mine networks and maintains the priority transmission capability of alarm messages over regular data.

[0101] In one graphical rollback configuration, the watchdog circuit of the twin visualization control terminal is connected to the graphics processor's reset pin or the driver's reset interface. A second memory stores 2D topology map resources and target coordinate annotation programs. A state where the terminal's video memory usage exceeds a preset threshold and the rendering thread is suspended serves as the trigger input for the watchdog circuit. The 2D topology map resources serve as the emergency display input, and the display interface serves as the top-level output for the screen. This configuration is suitable for scenarios involving 3D pipeline anomalies, video memory shortages, or graphics driver timeouts.

[0102] The system in this embodiment of the application uses a single alarm message to carry network identifier, redirection bit, peak time, and target coordinates, enabling communication queue control, graphics computing power redirection, and power cut-off control to share a common data source. The network identifier matches the target queue input of the exclusive transmission module, the redirection bit matches the dimensionality reduction control input of the rendering execution module, the peak time matches the rendering refresh priority mapping input, and the target coordinates match the warning zone positioning input and the power supply partition mapping input of the power cut-off module. This system can effectively reduce alarm message queuing during the accelerated phase of mine disaster trends to speed up physical power outages, reduce the probability of 3D rendering stuttering to ensure the continuity of situational awareness, provide monitoring personnel with a clear warning zone situation, and improve the response certainty of power cut-off corresponding to target coordinates.

[0103] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for data communication and visualization in predicting mine safety trends, applied to a mine safety monitoring system comprising a sensor array, measurement nodes, time-sensitive network switching equipment, a visualization terminal, and explosion-proof execution equipment, characterized in that, include: The measurement node acquires the real-time physical quantities of the sensor array using a preset first frequency, and calculates the trend approximation rate based on preset baseline values, limit values, anti-overflow constants and the first-order time derivative of the real-time physical quantities. When the trend approximation rate is greater than the preset threshold corresponding to the real-time physical quantity type, the measurement node switches to a preset second frequency to acquire the real-time physical quantity and generate an alarm message. The alarm message encapsulates the target coordinates, peak time, network identifier, and redirection bit. Specifically, when the absolute value of the first-order time derivative is greater than the preset minimum effective slope, the peak time is obtained by dividing the difference between the limit value and the real-time physical quantity by the first-order time derivative; otherwise, the peak time is set to a preset upper limit value. When the calculated peak time is less than or equal to a preset lower limit value, the peak time is clamped to the preset lower limit value. The time-sensitive network switching device identifies the network identifier, closes the transmission gate of the preset regular queue to truncate regular data, and establishes a dedicated channel to transmit the alarm message; Based on the redirection position, the visualization terminal reduces the rendering accuracy of the 3D model of the area outside the target coordinates, and performs 3D rendering of the warning area according to the rendering refresh priority mapped by the peak time, so as to present the disaster evolution margin and the physical isolation state of power outage. The shorter the peak time, the higher the rendering refresh priority it is mapped to. The explosion-proof actuator synchronously receives the alarm message through the dedicated channel and cuts off the power supply to the power supply zone or feeder circuit corresponding to the target coordinate.

2. The method for data communication and visualization in mine safety trend prediction according to claim 1, characterized in that, Before the measurement node acquires the real-time physical quantities of the sensing array using the preset first frequency, the method further includes: The visualization terminal preloads the full mesh constituting the three-dimensional model into the terminal's video memory and establishes a mapping relationship between the full mesh and spatial coordinates. The calculation node reads the preset baseline value and the preset limit value from a storage medium, and loads the preset anti-overflow constant for calculation.

3. The method for data communication and visualization in mine safety trend prediction according to claim 2, characterized in that, The calculation of the trend approximation rate based on preset baseline values, limit values, spillover prevention constants, and the first-order time derivative of the real-time physical quantities includes: When the real-time physical quantity is an upper limit physical quantity that represents an increase in danger due to a rise in its value, the difference between the real-time physical quantity and the preset baseline value is calculated within the calculation node and used as a fluctuation term. The difference between the preset limit value and the real-time physical quantity is added to the preset overflow prevention constant to obtain the margin value. When the margin value is less than or equal to zero, the margin value is clamped to the preset anti-overflow constant; When the direction of the first-order time derivative indicates that the real-time physical quantity is moving away from the preset limit value, the trend approximation rate is set to zero. Otherwise, the trend approximation rate is obtained by multiplying the ratio of the fluctuation term to the clamped margin value by the first time derivative of the real-time physical quantity.

4. The method for data communication and visualization in mine safety trend prediction according to claim 3, characterized in that, Within a preset period during which the measuring node switches from the preset first frequency to the preset second frequency, the method further includes: The measurement node calculates the variance of the sampling sequence at the frequency intersection point, performs smoothing processing on the real-time physical quantity accordingly, and calculates the first-order time derivative based on the smoothed real-time physical quantity. When the variance of the smoothed real-time physical quantity is higher than the preset variance threshold, or when the duration of the smoothing process exceeds the preset maximum suspension time, the calculation node performs calculations based on the highest historical effective derivative value instead of the first-order time derivative, or forcibly sets the peak time to a preset lower limit value.

5. The method for data communication and visualization in mine safety trend prediction according to claim 4, characterized in that, The generation of alarm messages by the measurement node includes: The measuring node writes the network identifier into the link frame header of the alarm message to mark the highest transmission priority; Write the redirection bit into the payload header of the alarm message, and write the peak time, the target coordinates, and the sampled sequence summary of the alarm trigger time into the payload of the alarm message.

6. The method for data communication and visualization in mine safety trend prediction according to claim 5, characterized in that, The step of the time-sensitive network switching device identifying the network identifier, closing the transmission gate of the preset regular queue to truncate regular data, and establishing a dedicated channel to transmit the alarm message includes: The time-sensitive network switching device truncates the regular data based on the network identifier in the link frame header; A frame preemption mechanism is adopted to suspend frame transmission in the preset regular queue in order to release bandwidth; The target queue with the highest transmission priority is assigned exclusive bandwidth to send the alarm message.

7. The method for data communication and visualization in mine safety trend prediction according to claim 6, characterized in that, In the process of sending the alarm message with dedicated bandwidth, the method further includes: When the buffer depth of the target queue reaches a preset congestion threshold due to uplink bandwidth bottleneck, the time-sensitive network switching device returns a suppression frame to the measurement node. Based on the suppression frame, the measurement node reduces or discards the sampling sequence summary in the payload in the subsequently generated alarm message, while maintaining the integrity of the network identifier, the redirection bit, the target coordinates, and the peak time.

8. The method for data communication and visualization in mine safety trend prediction according to claim 7, characterized in that, The process involves the visualization terminal reducing the rendering precision of the 3D model in the region outside the target coordinates based on the redirection bit, thereby freeing up graphics processor computing resources and improving the rendering capability of the warning zone corresponding to the target coordinates. The 3D rendering of the warning zone is performed according to the rendering refresh priority mapped to the peak time, including: Based on the redirection bit, the geometric phase of the 3D model located outside the target coordinates is simplified to include distant mesh calculations, or complex shaders and post-processing effects of the 3D model in the outer region are disabled to reduce the shading calculation load on the graphics processor and to avoid triggering the unloading and reloading of the full mesh. The graphics processor's rendering pipeline is focused onto the warning zone corresponding to the target coordinates; Using the peak time as input, a preset rendering refresh priority mapping table is queried to obtain the corresponding rendering refresh priority, and the rendering pipeline is invoked to perform a high-priority thermal rendering task on the warning zone.

9. The method for data communication and visualization in mine safety trend prediction according to claim 8, characterized in that, During the execution of the high-priority thermal rendering task on the warning area, the method further includes: When the terminal video memory occupancy rate exceeds a preset occupancy threshold within a preset period, causing the rendering thread to be suspended, a watchdog reset is triggered to reset the 3D rendering pipeline containing the rendering thread. The vector interface is invoked to draw a two-dimensional topology map containing the target coordinates on the top layer of the screen.

10. A mine safety trend prediction data communication and visualization system, characterized in that, The system is used to implement the mine safety trend prediction data communication and visualization method as described in any one of claims 1 to 9, the system comprising: The physical quantity measurement module is used to acquire the real-time physical quantities of the sensor array using a preset first frequency, and to calculate the trend approximation rate based on preset baseline values, limit values, anti-overflow constants and the first-order time derivative of the real-time physical quantities. An alarm generation module is used to switch to a preset second frequency to acquire the real-time physical quantity and generate an alarm message when the trend approximation rate is greater than a preset threshold corresponding to the real-time physical quantity type. The alarm message encapsulates the target coordinates, peak time, network identifier, and redirection bit. Specifically, when the absolute value of the first-order time derivative is greater than a preset minimum effective slope, the peak time is obtained by dividing the difference between the limit value and the real-time physical quantity by the first-order time derivative; otherwise, the peak time is set to a preset upper limit value. When the calculated peak time is less than or equal to a preset lower limit value, the peak time is clamped to the preset lower limit value. The dedicated transmission module is used to identify the network identifier, close the transmission gate of the preset regular queue to truncate regular data, and establish a dedicated channel to transmit the alarm message; The rendering execution module is used to reduce the rendering accuracy of the three-dimensional model of the region outside the target coordinates based on the redirection bit, and to perform three-dimensional rendering of the warning area according to the rendering refresh priority mapped by the peak time, so as to present the disaster evolution margin and the power outage physical isolation state. The shorter the peak time, the higher the rendering refresh priority it is mapped to. The power supply interruption module is used to synchronously receive the alarm message through the exclusive channel and cut off the power supply to the power supply zone or feeder circuit corresponding to the target coordinate.

Images