Safety early warning system for underground gas pipe network

Abstract

The invention relates to the technical field of urban public safety, in particular to an underground gas pipe network safety pre-warning system, which comprises an acquisition module, a monitoring module, a monitoring module and a control module, and is characterized in that the acquisition module is used for arranging sensing nodes at key positions of a gas pipeline body and a surrounding environment to form a distributed monitoring network; the method comprises the following steps: synchronously acquiring mechanical deformation, methane concentration and mechanical vibration data of a pipeline through a distributed monitoring network to obtain a multi-modal sensing data set; the fusion module is used for performing humidity and temperature interference correction on the methane concentration data in the multi-modal sensing data set to obtain corrected methane concentration data; and performing fusion processing on the mechanical deformation data, the mechanical vibration data and the corrected methane concentration data to obtain an initial pipeline safety state evaluation index. Through multi-source data fusion, environmental interference correction and spatial domain calibration, transmission and dynamic analysis are combined, and precise early warning and efficient linkage disposal of the safety state of the gas pipe network are achieved.

Description

Technical Field

[0001] This invention relates to the field of urban public safety technology, and more specifically, to an underground gas pipeline safety early warning system. Background Technology

[0002] Underground gas pipelines are a vital urban energy infrastructure, facing complex threats to their safe operation, such as soil corrosion and third-party construction. Achieving accurate and reliable safety early warning systems is a current research focus. However, existing monitoring systems have significant shortcomings in data processing, hindering the improvement of their early warning effectiveness.

[0003] Existing technologies largely rely on independent threshold judgments for single parameters such as methane concentration and vibration, resulting in limited information. First, methane concentration sensor readings are easily affected by ambient temperature and humidity, causing drift, while vibration data struggles to distinguish between hazardous construction vibrations and normal traffic vibrations, potentially leading to frequent false alarms and missed alarms. For example, in subway construction scenarios, vibrations from large machinery can cause stress concentration in pipelines, while changes in groundwater levels caused by construction can increase ambient humidity. Traditional systems may misinterpret sensor drift as gas leaks due to humidity interference (false alarms), or completely ignore potential leak risks due to the inability to detect mechanical deformation (missed alarms), until the pipeline is actually damaged and leaks, at which point the warning has lost its meaning.

[0004] Secondly, analysis based on single-point data ignores the integrity of pipelines as linear infrastructure and cannot perform regional weighting and calibration of risks based on the spatial topology of the sensor network, which may lead to inaccurate early warning positioning and one-sided level assessment. In the above case, if the construction vibration source is located at monitoring point A, but its main risk may affect the area where monitoring point B is located downstream, the traditional single-point assessment mode cannot achieve accurate positioning and early warning of such risk transmission.

[0005] Finally, relying on a single communication mode may cause data interruption or delay in underground environments with poor signal, preventing the aforementioned critical early warning information from being delivered to the monitoring center in a timely manner and delaying emergency response. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide an underground gas pipeline safety early warning system. By integrating multi-source data fusion, environmental interference correction and spatial domain calibration, combined with transmission and dynamic analysis, the system achieves accurate early warning and efficient linkage response to the safety status of the gas pipeline network.

[0007] To solve the above-mentioned technical problems, the basic concept of the technical solution adopted by the present invention is as follows: Firstly, an underground gas pipeline network safety early warning system includes: The data acquisition module is used to deploy sensor nodes at key locations on the gas pipeline and its surrounding environment to form a distributed monitoring network; the distributed monitoring network synchronously collects data on the pipeline's mechanical deformation, methane concentration and mechanical vibration to obtain a multimodal sensing data set. The fusion module is used to correct the methane concentration data in the multimodal sensing dataset for humidity and temperature interference to obtain corrected methane concentration data; it also fuses the mechanical deformation data, mechanical vibration data and corrected methane concentration data to obtain the initial pipeline safety status assessment index. The calibration module receives the initial pipeline safety status assessment index, delineates the corresponding polygonal monitoring area based on the topology of the distributed monitoring network, and performs grid division to obtain the area weight coefficient of each grid. Based on the weight coefficient, a dynamic adjustment coefficient is obtained through weighted calculation, and the initial pipeline safety status assessment index is spatially calibrated using the dynamic adjustment coefficient to obtain a comprehensive pipeline safety status assessment index. The transmission module is used to receive comprehensive pipeline safety status assessment indicators, transmit them through a dual-mode communication link, and use edge computing nodes to relay and cache the transmitted data, successfully uploading the indicators to the monitoring center to obtain pipeline safety status data located at the monitoring center. The early warning module is used to perform dynamic risk analysis on pipeline safety status data and obtain analysis results. When the analysis results exceed the preset threshold, an early warning is automatically triggered and linkage instructions are sent to the gas operation and maintenance platform, construction unit and community emergency system to achieve closed-loop management.

[0008] Secondly, a control method for an underground gas pipeline network safety early warning system includes the following steps: Sensor nodes are deployed at key locations on the gas pipeline and its surrounding environment to form a distributed monitoring network; the mechanical deformation, methane concentration and mechanical vibration data of the pipeline are collected synchronously through the distributed monitoring network to obtain a multimodal sensing data set; Humidity and temperature interference corrections were applied to the methane concentration data in the multimodal sensing dataset to obtain corrected methane concentration data. The mechanical deformation data, mechanical vibration data and corrected methane concentration data were then fused together to obtain the initial pipeline safety status assessment index. The system receives the initial pipeline safety status assessment index, delineates the corresponding polygonal monitoring area based on the topology of the distributed monitoring network, and performs grid division to obtain the area weight coefficient of each grid. Based on this weight coefficient, a dynamic adjustment coefficient is obtained through weighted calculation, and the initial pipeline safety status assessment index is spatially calibrated using the dynamic adjustment coefficient to obtain a comprehensive pipeline safety status assessment index. The system receives comprehensive pipeline safety status assessment indicators, transmits them through a dual-mode communication link, and uses edge computing nodes to relay and cache the transmitted data. The indicators are then successfully uploaded to the monitoring center to obtain pipeline safety status data located at the monitoring center. Dynamic risk analysis is performed on pipeline safety status data to obtain analysis results. When the analysis results exceed the preset threshold, an early warning is automatically triggered and linkage instructions are sent to the gas operation and maintenance platform, construction unit and community emergency system simultaneously to achieve closed-loop management.

[0009] By adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art.

[0010] By deploying sensor nodes at key locations within the gas pipeline and its surrounding environment to form a distributed monitoring network, comprehensive coverage of critical pipeline areas can be achieved. This network simultaneously collects multimodal sensing data sets, integrating pipeline status information from different dimensions. Humidity and temperature interference corrections are applied to methane concentration data within the multimodal sensing data set to reduce the impact of environmental factors on the accuracy of methane concentration data. Initial pipeline safety status assessment indicators are obtained through fusion processing, integrating the value of different data types and allowing the initial indicators to more comprehensively reflect the pipeline's safety condition. Based on the topology of the distributed monitoring network, corresponding polygonal monitoring areas are delineated and gridded, ensuring that the assessment scope matches the actual pipeline distribution and monitoring node layout, clearly defining the assessment objects for each specific area. The area weight coefficients of each grid are obtained, and a dynamic adjustment coefficient is calculated through weighted summation. This coefficient is then used to spatially calibrate the initial pipeline safety status assessment indicators to obtain a comprehensive pipeline safety status assessment index. The system can consider the impact of spatial differences in different grid areas on safety assessments, avoiding spatial dimension assessment biases and making comprehensive indicators more consistent with the actual spatial safety situation of pipelines. Transmission via dual-mode communication links ensures data transmission stability. Utilizing edge computing nodes for data relay and caching alleviates the pressure of direct data transmission, reduces the probability of data loss or delay, and ensures that comprehensive pipeline safety status assessment indicators are successfully uploaded to the monitoring center, enabling the center to obtain complete pipeline safety status data in a timely manner. Dynamic risk analysis of pipeline safety status data yields analysis results, allowing for continuous tracking of changes in pipeline safety status and timely understanding of risk development trends. Automatic warnings are triggered when analysis results exceed preset thresholds, quickly indicating the existence of risks and avoiding delays in risk response. Simultaneous linkage commands are sent to the gas operation and maintenance platform, construction units, and community emergency systems to achieve closed-loop management, integrating multiple response capabilities and improving the efficiency of risk response and handling.

[0011] The specific embodiments of the present invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0012] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. Some specific embodiments of this application will be described in detail below with reference to the accompanying drawings in an exemplary and non-limiting manner. The same reference numerals in the drawings designate the same or similar parts or components. Those skilled in the art should understand that these drawings are not necessarily drawn to scale. In the drawings: Figure 1 This is a schematic diagram of the underground gas pipeline safety early warning system of the present invention.

[0013] Figure 2 This is a schematic diagram of the control method of the underground gas pipeline safety early warning system of the present invention.

[0014] It should be noted that these accompanying drawings and textual descriptions are not intended to limit the scope of the invention in any way, but rather to illustrate the concept of the invention to those skilled in the art by referring to specific embodiments. The elements in the drawings are schematic and not drawn to scale. Detailed Implementation

[0015] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative effort should fall within the scope of protection of the present application.

[0016] The following embodiments of this application use an underground gas pipeline network safety early warning system as an example to illustrate the solution of this application in detail. However, this embodiment does not limit the scope of protection of this application.

[0017] like Figure 1 As shown, the present invention provides a safety early warning system for underground gas pipeline networks, comprising: The acquisition module 11 is used to deploy sensor nodes at key locations on the gas pipeline body and its surrounding environment to form a distributed monitoring network; the mechanical deformation, methane concentration and mechanical vibration data of the pipeline are collected synchronously through the distributed monitoring network to obtain a multimodal sensing data set; The fusion module 12 is used to correct the methane concentration data in the multimodal sensing dataset for humidity and temperature interference to obtain corrected methane concentration data; and to fuse the mechanical deformation data, mechanical vibration data and corrected methane concentration data to obtain the initial pipeline safety status assessment index. The calibration module 13 is used to receive the initial pipeline safety status assessment index, delineate the corresponding polygonal monitoring area based on the topology of the distributed monitoring network, and perform grid division to obtain the area weight coefficient of each grid; based on the weight coefficient, the dynamic adjustment coefficient is obtained through weighted calculation, and the initial pipeline safety status assessment index is spatially calibrated using the dynamic adjustment coefficient to obtain the comprehensive pipeline safety status assessment index. The transmission module 14 is used to receive comprehensive pipeline safety status assessment indicators, transmit them through a dual-mode communication link, and use edge computing nodes to relay and cache the transmitted data, successfully uploading the indicators to the monitoring center to obtain pipeline safety status data located in the monitoring center. The early warning module 15 is used to perform dynamic risk analysis on pipeline safety status data and obtain analysis results. When the analysis results exceed the preset threshold, an early warning is automatically triggered and linkage instructions are sent to the gas operation and maintenance platform, construction unit and community emergency system to achieve closed-loop management.

[0018] In this embodiment of the invention, by deploying sensing nodes at key locations on the gas pipeline body and its surrounding environment to form a distributed monitoring network, comprehensive coverage of different areas of the pipeline can be achieved. The multimodal sensing data set can simultaneously capture information related to pipeline structural status, gas leakage, and external interference factors. Humidity and temperature interference corrections are applied to the methane concentration data in the multimodal sensing data set to eliminate the influence of environmental factors on the methane concentration detection results. The initial pipeline safety status assessment index is obtained by fusing mechanical deformation data, mechanical vibration data, and the corrected methane concentration data. This integrates pipeline safety information reflected by different types of data, enabling the initial assessment index to comprehensively reflect pipeline structural stability, gas leakage risk, and external disturbances. Based on the topology of the distributed monitoring network, corresponding polygonal monitoring areas are delineated and gridded, allowing for refinement of the monitoring range based on the actual distribution of monitoring nodes. By obtaining the area weight coefficient of each grid and then calculating a dynamic adjustment coefficient based on the weight coefficient, the monitoring range can be adjusted according to the different grid areas within the overall monitoring network. The importance of each pipeline in the network is assigned a corresponding weight. A comprehensive pipeline safety status assessment index is obtained by spatially calibrating the initial pipeline safety status assessment index using dynamic adjustment coefficients. This allows the assessment index to adapt to the monitoring needs of different spatial areas and better match the spatial distribution characteristics of actual monitoring scenarios. The comprehensive pipeline safety status assessment index is transmitted through a dual-mode communication link, leveraging the complementarity of the two communication methods to improve data transmission reliability. Using edge computing nodes to relay and cache transmitted data reduces latency and data loss during long-distance transmission. Dynamic risk analysis of pipeline safety status data allows for real-time tracking of changes in pipeline safety status and timely detection of abnormal trends. When the analysis results exceed a preset threshold, an early warning is automatically triggered, quickly initiating the safety alert process. Simultaneous sending of linkage instructions to the gas operation and maintenance platform, construction units, and community emergency systems allows multiple relevant entities to simultaneously obtain early warning information and handling requirements, enabling rapid collaborative safety handling and forming a closed-loop management system from early warning to handling, thus improving the efficiency of safety incident response.

[0019] In the underground gas pipeline safety early warning system described in this embodiment of the invention, the above-mentioned module 11 deploys sensing nodes at key locations on the gas pipeline body and its surrounding environment to form a distributed monitoring network; the distributed monitoring network synchronously collects data on the pipeline's mechanical deformation, methane concentration, and mechanical vibration to obtain a multimodal sensing data set, including: Step 111: Based on the gas pipeline's laying path, surrounding environmental risk level, and historical accident data, obtain the optimal sensor node topology covering the pipeline itself and adjacent key areas. Specifically, this includes: first, collecting complete laying path information for the gas pipeline, including the actual route, burial depth, pipe diameter, pipe material type, and the relative position of the pipeline with surrounding buildings and other underground pipelines; then, conducting a risk level assessment of the surrounding environment, classifying high-risk areas such as areas with active third-party construction, areas with high traffic volume, and areas with highly corrosive soil; medium-risk areas such as pipeline sections around residential areas; and low-risk areas such as pipeline sections in open suburban areas; then retrieving the gas pipeline's data... Historical accident data of the pipeline over the past 5 to 10 years is used to statistically analyze the location, cause (e.g., corrosion leaks, construction damage), impact range, and losses of each accident. Then, the laying path information, surrounding environmental risk level assessment results, and historical accident data are imported into the data analysis system. Using spatial interpolation combined with risk weight calculation, the deployment density of sensor nodes in different areas is determined. In high-risk areas and areas with frequent historical accidents, one node is deployed every 20 to 30 meters; in medium-risk areas, one node is deployed every 30 to 40 meters; and in low-risk areas, one node is deployed every 40 to 50 meters. Finally, the optimal sensor node topology covering the pipeline body and adjacent key areas is generated.

[0020] Step 112: Based on the aforementioned topology, deploy micro-displacement sensors, methane concentration sensors, and vibration sensors at corresponding spatial coordinate locations to form a distributed monitoring network with spatial correlation. Specifically, this includes: First, based on the optimal sensor node topology determined in Step 111, using professional surveying equipment such as a GPS locator or total station, determine the precise spatial coordinates of each sensor node, including longitude, latitude, and altitude; then, according to the spatial coordinates of each sensor node, conduct underground excavation at the corresponding location, with the excavation depth matching the burial depth of the gas pipeline to ensure the sensors can effectively monitor the pipeline status; then, at each excavation location... After excavation, micro-displacement sensors, methane concentration sensors, and vibration sensors are installed at the sensor node locations. The micro-displacement sensors are fixed close to the outer wall of the gas pipeline to monitor the pipeline's mechanical deformation. The methane concentration sensors are placed in the soil around the pipeline to monitor gas leaks. The vibration sensors are fixed to the soil or ground structure near the pipeline to monitor vibration signals. Then, the spatial coordinate information corresponding to each sensor is recorded, and spatial correlation is established between the three sensors of each sensor node and the sensors of adjacent sensor nodes. Information such as the distance and relative orientation between adjacent nodes is recorded, ultimately forming a distributed monitoring network with spatial correlation.

[0021] Step 113: Through the distributed monitoring network, each sensor is synchronously triggered to collect data according to a unified time reference, acquiring pipeline mechanical deformation data, methane concentration data, and mechanical vibration data respectively. Specifically, this includes: first, establishing a standard time synchronization system at the gas pipeline monitoring center, setting the system's time as the unified time reference for the entire distributed monitoring network; then, transmitting the unified time reference signal from the monitoring center to each sensor node via a LoRa and NB-IoT dual-mode communication link, calibrating the local clock of each sensor node to ensure that the time of all sensor nodes is consistent with the time of the monitoring center, with the time error controlled within milliseconds. Within the specified range; then, based on the safety monitoring requirements of the gas pipeline network, a data acquisition cycle is set, such as data acquisition every 5 minutes; when the preset acquisition time point is reached, the monitoring center sends a synchronous acquisition command to all sensor nodes through a dual-mode communication link; after receiving the command, each sensor node simultaneously triggers the micro-displacement sensor to acquire the mechanical deformation data of the gas pipeline, triggers the methane concentration sensor to acquire the methane concentration data around the pipeline, and triggers the vibration sensor to acquire the mechanical vibration data of the surrounding environment; after each sensor completes its acquisition, the acquired raw data is temporarily stored in the local data cache module of the sensor node, awaiting subsequent data processing operations.

[0022] Step 114 involves performing timestamp alignment, outlier removal, and dimensional normalization on the mechanical deformation data, methane concentration data, and mechanical vibration data, integrating them to generate a multimodal sensing data set with a consistent spatiotemporal reference. Specifically, this includes: firstly, extracting the collected mechanical deformation data, methane concentration data, and mechanical vibration data from the local data cache module of each sensor node, and simultaneously extracting the collection timestamp corresponding to each data point; then, verifying and adjusting the timestamps of all data according to the unified time reference set by the monitoring center, correcting data with time deviations according to the unified time reference, and achieving timestamp alignment of data from different sensors and different sensor nodes; subsequently, for spatial data gaps in the timestamp-aligned data set, using an inverse distance weighted spatial interpolation algorithm to supplement the data. That is, for each gap location, based on the similar data of the 6 to 8 nearest valid sensor nodes, different weights are assigned according to the distance between the node and the gap location, with closer distances resulting in greater weights. The supplementary data for the gap location is obtained through weighted calculation, making the data more continuous in spatial distribution.

[0023] Next, the k-nearest neighbor spatial anomaly detection method is combined with the original anomaly judgment rules to process anomalies. For each data point, the five nearest neighboring sensor nodes in space are found to have the same type of data at the same time stamp. The average deviation between this data and the neighboring data is calculated. If the deviation exceeds a preset threshold and simultaneously meets the conditions of exceeding the normal value range or fluctuating drastically in a short period of time without reasonable external causes, it is judged as a spatial anomaly. Such anomalies are removed from the data set along with the anomalies identified by the original rules. Then, for the dimensional differences of different types of data, a linear normalization method is used to convert mechanical deformation data into dimensionless values ​​of 0 to 1, methane concentration data into dimensionless values ​​of 0 to 1, and mechanical vibration data into dimensionless values ​​of 0 to 1.

[0024] Finally, according to the timestamp order and the spatial coordinates of the sensor nodes, the processed mechanical deformation data, methane concentration data, mechanical vibration data, and spatially interpolated supplementary data are integrated to form a dataset in which each data point contains data type, acquisition time, spatial coordinates, and dimensionless value. In the end, a multimodal sensing data set with a consistent spatiotemporal reference, continuous spatial distribution, and accurate removal of outliers is generated.

[0025] In this embodiment of the invention, the optimal sensor node topology is determined based on the gas pipeline's laying path, the surrounding environmental risk level, and historical accident data. This allows the sensor node layout to fully match the actual pipeline route and risk distribution characteristics. By combining historical accident data, node coverage in areas where safety issues have occurred can be specifically strengthened, making the monitoring network layout more aligned with actual safety needs. Deploying micro-displacement sensors, methane concentration sensors, and vibration sensors at corresponding spatial coordinates according to the optimal topology ensures that the installation positions of various sensors accurately correspond to the spatial characteristics of the pipeline and its surrounding environment. This forms a distributed monitoring network with spatial correlation, allowing data collected by different sensors to be bound to specific spatial locations, facilitating rapid location of the data source area during subsequent data analysis. Synchronous triggering of data collection by each sensor according to a unified time reference enables… To ensure pipeline-related data are acquired simultaneously by micro-displacement sensors, methane concentration sensors, and vibration sensors, timestamp alignment of mechanical deformation data, methane concentration data, and mechanical vibration data further strengthens the temporal correlation of different data types, ensuring all data are based on a consistent time reference, facilitating subsequent integration and analysis. Outlier removal filters out invalid data generated by sensor malfunctions, transient external interference, and other factors, reducing the interference of outliers on subsequent evaluation results and improving data quality. Dimensional normalization eliminates analytical obstacles caused by dimensional differences in different data types, allowing data from different dimensions such as mechanical deformation, methane concentration, and mechanical vibration to be integrated within the same analytical framework. Ultimately, this generates a multimodal sensing data set with a consistent spatiotemporal reference, providing a unified, high-quality data foundation for subsequent data fusion processing.

[0026] In the underground gas pipeline safety early warning system described in this embodiment of the invention, module 12 corrects the methane concentration data in the multimodal sensing data set for humidity and temperature interference to obtain corrected methane concentration data; it then fuses the mechanical deformation data, mechanical vibration data, and the corrected methane concentration data to obtain initial pipeline safety status assessment indicators, including: Step 121 involves separating methane concentration data from the multimodal sensing data set and simultaneously reading the ambient temperature and humidity data collected by the temperature and humidity sensor built into the methane concentration sensor within the corresponding time period. Specifically, this includes: firstly, classifying and organizing the multimodal sensing data set, where each data entry has a data type label, a collection timestamp, and a sensor node number; then, filtering data entries labeled as methane concentration through the data management system to separate all methane concentration data, while recording the collection timestamp and sensor node number corresponding to each methane concentration data entry; next, locating the corresponding methane concentration sensor based on the sensor node number of the separated methane concentration data, calling the stored data of the temperature and humidity module built into the sensor, and filtering the ambient temperature and humidity data within the same time period and under the same sensor node number according to the collection timestamp of the methane concentration data, ensuring that each methane concentration data entry can be matched with temperature and humidity data with completely consistent collection times, forming a one-to-one associated dataset of methane concentration data-temperature and humidity data-timestamp-node number.

[0027] Step 122: Based on the ambient temperature and humidity data, query the preset temperature-humidity-concentration compensation relationship mapping table to obtain the corresponding concentration compensation coefficient set. The concentration compensation coefficient set includes slope adjustment parameters and intercept compensation parameters corresponding to different temperature and humidity combinations. Specifically, when setting the temperature-humidity-concentration compensation relationship mapping table, first conduct laboratory simulation tests to determine the temperature test range as -10℃ to 40℃ and the humidity test range as 20%RH to 90%RH. Within this range, divide a temperature gradient every 2℃ and a humidity gradient every 5%RH. Each combination of temperature gradient and humidity gradient constitutes an independent test condition. Under each condition, introduce standard methane gas with a known concentration into the methane concentration sensor through a standard gas generator. After the sensor reading stabilizes, record the measured value. Compare the measured value with the standard gas concentration value to calculate the deviation. Based on the deviation, determine the slope adjustment parameters and intercept compensation parameters corresponding to the condition through data fitting. After completing all condition tests, collect a preliminary parameter set.

[0028] Next, on-site testing and parameter correction were conducted. Monitoring points for gas pipelines covering different actual environments, such as high temperature and high humidity, low temperature and low humidity, and normal temperature and normal humidity, were selected. At each monitoring point, a methane concentration sensor, a built-in temperature and humidity sensor, and a gas chromatograph were deployed simultaneously to continuously collect environmental temperature and humidity data, measured concentration data from the methane concentration sensor, and standard concentration data detected by the gas chromatograph. The temperature and humidity data collected on-site were mapped to similar operating conditions simulated in the laboratory. The differences between the laboratory parameters and the parameters required for deviation from the on-site measurements were compared, and the preliminary laboratory parameters were fine-tuned to ensure that the parameters could adapt to the influence of temperature and humidity in the actual environment.

[0029] Finally, all the operating parameters that have been simulated in the laboratory and corrected on-site are compiled and summarized according to the correspondence between temperature value, humidity value, corresponding slope adjustment parameter, and corresponding intercept compensation parameter, forming a temperature-humidity-concentration compensation relationship mapping table covering all operating conditions, thus completing the pre-setting of the table.

[0030] First, the pre-set temperature-humidity-concentration compensation relationship mapping table is called, and each piece of environmental temperature and humidity data (including specific temperature and humidity values) in the associated dataset obtained in step 121 is input into the mapping table query system. The query system locates the corresponding temperature range in the pre-set table based on the input temperature value, and locates the corresponding humidity range based on the input humidity value, thus determining the specific operating condition range to which the current temperature and humidity combination belongs. The slope adjustment parameter and intercept compensation parameter stored in the pre-set table are extracted from the pre-set table, and these two parameters are integrated to form a set of concentration compensation coefficients for the current temperature and humidity conditions.

[0031] Step 123: Based on the ambient temperature and humidity values ​​corresponding to the methane concentration data, select the appropriate slope adjustment parameter and intercept compensation parameter from the concentration compensation coefficient set; multiply the slope adjustment parameter by the corresponding methane concentration data to obtain a preliminary correction value; add the preliminary correction value to the selected intercept compensation parameter to obtain the methane concentration data after environmental interference correction. Specifically, this includes: firstly, extracting a single methane concentration data and its corresponding ambient temperature and humidity values ​​from the associated dataset formed in step 121; inputting the temperature and humidity values ​​into the concentration compensation coefficient matching module; and searching for the temperature and humidity combination that is completely consistent with or has the smallest deviation from the temperature and humidity values ​​in the concentration compensation coefficient set obtained in step 122. The slope adjustment parameter and intercept compensation parameter are selected. Then, the data correction calculation process is started. The selected slope adjustment parameter is multiplied with the currently processed methane concentration data to obtain a preliminary correction value. For example, if the slope adjustment parameter is 1.05 and the methane concentration data is 8%LEL (where the methane concentration data is in %LEL), then the preliminary correction value is 8.4%LEL. Next, the preliminary correction value is added with the selected intercept compensation parameter. If the intercept compensation parameter is -0.2%LEL, then the final methane concentration data after environmental interference correction is 8.2%LEL. The above process is followed to process all methane concentration data in the associated dataset to obtain the complete corrected methane concentration dataset.

[0032] Step 124: Based on the mechanical deformation data and mechanical vibration data separated from the multimodal sensing data set, time-domain and frequency-domain feature extraction processes are performed respectively to obtain feature vectors representing the pipeline stress state and feature vectors representing the intensity of external disturbances. Specifically, this includes: firstly, selecting data entries labeled as mechanical deformation and mechanical vibration from the multimodal sensing data set, separating the mechanical deformation data and mechanical vibration data respectively, and sorting them according to the acquisition timestamp; performing time-domain feature extraction on the mechanical deformation data, calculating the maximum deformation value, minimum deformation value, deformation rate, and mean deformation value within each time window to obtain time-domain feature parameters; and then converting the mechanical deformation data to the frequency domain through Fourier transform to extract... The characteristic frequency, peak frequency, and corresponding energy percentage in the frequency domain are used to form a feature vector characterizing the pipeline stress state by arranging the time-domain and frequency-domain feature parameters in a preset order. The vector dimension is determined according to the number of extracted feature parameters. For example, when the four parameters of maximum deformation value, deformation rate, characteristic frequency, and peak frequency are included, the feature vector is [maximum deformation value, deformation rate, characteristic frequency, peak frequency]. The same processing method is used for mechanical vibration data. The vibration peak value, root mean square value, and vibration duration are extracted in the time domain, and the main frequency band, main frequency band energy, and secondary main frequency band frequency are extracted in the frequency domain. These parameters are integrated in order to form a feature vector characterizing the intensity of external disturbance.

[0033] Step 125 involves fusing the corrected methane concentration data with the stress state feature vector and disturbance intensity feature vector through multi-source data fusion to obtain initial pipeline safety status assessment indicators. Specifically, this includes: firstly, standardizing the corrected methane concentration data to convert its numerical range to a dimensionless interval of 0 to 1, consistent with the stress state feature vector and disturbance intensity feature vector; then, introducing multidimensional scaling analysis to optimize the dimensions of the standardized multi-source data, calculating the Euclidean distance between the corrected methane concentration data, the parameters of each dimension of the stress state feature vector, and the parameters of each dimension of the disturbance intensity feature vector, constructing a distance matrix for the multi-source data, and mapping the distance matrix to a low-dimensional space (e.g., 2 to 3 dimensions) using a multidimensional scaling analysis algorithm, preserving the core correlations between the data, and simultaneously selecting key parameters that contribute significantly to the pipeline safety status assessment, eliminating redundant dimension parameters, and forming a simplified and complete core data set.

[0034] Then, combining the requirements of gas pipeline network safety assessment, the contribution results of key parameters obtained from multidimensional scaling analysis are combined with expert review and historical data verification to determine the weight coefficients corresponding to each key parameter. The key parameter with higher contribution has a larger weight coefficient, ensuring that the sum of the weight coefficients of all key parameters is 1. Next, the key parameters selected by multidimensional scaling analysis are multiplied by their corresponding weight coefficients to obtain the weighted values ​​of each key parameter. For example, the weighted values ​​of key parameters corresponding to the corrected methane concentration data, the weighted values ​​of key parameters related to stress state, and the weighted values ​​of key parameters related to disturbance intensity are all calculated. Finally, the weighted values ​​of all key parameters are summed to obtain a single comprehensive value.

[0035] Finally, the comprehensive value is defined as the initial pipeline safety status assessment index. The larger the index value, the higher the current safety risk of the pipeline. At the same time, the complete calculation basis of the index is recorded, including the standardization process, the distance matrix construction and dimensionality reduction process of multidimensional scale analysis, the key parameter screening results, the basis for determining the weight coefficients and the weighting of each key parameter, forming a complete data archive of the initial assessment index including the multidimensional scale analysis.

[0036] In this embodiment of the invention, methane concentration data is separated from the multimodal sensing data set, enabling accurate extraction of key data of a single type. Simultaneous reading of ambient temperature and humidity data from the built-in temperature and humidity sensor of the methane concentration sensor during the corresponding time period ensures a complete match between the temperature and humidity data and the methane concentration data in terms of acquisition time, establishing a direct spatiotemporal correlation between the two. A pre-set temperature-humidity-concentration compensation relationship mapping table is queried based on the ambient temperature and humidity data, allowing direct access to a pre-built mature compensation parameter system, improving data processing efficiency. The concentration compensation coefficient set includes slope adjustment parameters and intercept compensation parameters corresponding to different temperature and humidity combinations, providing differentiated compensation basis for diverse temperature and humidity scenarios, making the compensation process more closely aligned with specific environmental conditions. Matching and selecting the corresponding slope adjustment parameters and intercept compensation parameters according to the temperature and humidity values ​​corresponding to the methane concentration data achieves accurate correspondence between the compensation parameters and the actual environmental state. A preliminary correction value is obtained by multiplying the slope adjustment parameter by the methane concentration data, and then the intercept compensation parameter is superimposed to complete the correction. A quantitative measurement is used. The calculation method achieves precise cancellation of temperature and humidity interference, effectively eliminating methane concentration data drift caused by environmental factors, making the corrected methane concentration data more consistent with the actual gas leakage situation and improving data accuracy. Time-domain and frequency-domain feature extraction processing is performed on mechanical deformation data and mechanical vibration data respectively, separating key information characterizing the core state of the pipeline from the original data, such as deformation peak value and vibration amplitude in the time domain, and characteristic frequencies in the frequency domain. The extracted key information is transformed into feature vectors characterizing the pipeline stress state and feature vectors characterizing the intensity of external disturbances, transforming complex and redundant original data into a structured and targeted data format. Multi-source data fusion of the corrected methane concentration data, stress state feature vectors, and disturbance intensity feature vectors integrates key safety information from three dimensions: gas leakage, pipeline structure, and external disturbances. Through fusion processing, data of different types and with different representational meanings are transformed into a unified initial pipeline safety status assessment index, enabling the assessment index to comprehensively reflect the multi-faceted safety status of the pipeline.

[0037] In the underground gas pipeline safety early warning system described in this embodiment of the invention, module 13 receives the initial pipeline safety status assessment index, delineates the corresponding polygonal monitoring area based on the topology of the distributed monitoring network, and performs grid division to obtain the area weight coefficient of each grid; based on the weight coefficient, a dynamic adjustment coefficient is obtained through weighted calculation, and the initial pipeline safety status assessment index is spatially calibrated using the dynamic adjustment coefficient to obtain a comprehensive pipeline safety status assessment index, including: Step 131 involves receiving the initial pipeline safety status assessment index and obtaining the optimal sensor node topology constituting the distributed monitoring network. Specifically, this includes: firstly, receiving the initial pipeline safety status assessment index through the system's internal data transmission interface. This index includes the sensor node number, acquisition timestamp, and specific assessment value corresponding to each data point, and temporarily storing this information in a dedicated data cache unit; then, calling the system's preset distributed monitoring network configuration database to extract the optimal sensor node topology information constituting the network. This information includes the device number of all sensor nodes, the latitude, longitude, and altitude coordinates of their deployment locations, sensor types (micro-displacement sensors, methane concentration sensors, vibration sensors), the connection relationships between nodes, and the deployment density distribution. This ensures that the obtained topology is completely consistent with the actual monitoring network, providing a precise basis for the subsequent delineation of the monitoring area.

[0038] Step 132: Based on the optimal sensor node topology, extract the spatial coordinates of all key sensor nodes to form a set of key node spatial coordinates; for the set of key node spatial coordinates, using the point with the smallest ordinate as the reference point, perform a convex hull algorithm to calculate the ordered boundary point sequence constituting the minimum convex boundary; connect each boundary point sequentially according to the order of the boundary point sequence to form the minimum circumscribed convex polygon, and determine this polygon as the corresponding polygon monitoring area. Specifically, this includes: firstly, from the optimal sensor node topology obtained in step 131, select the key sensor nodes that directly participate in pipeline safety status monitoring, exclude backup nodes and redundant communication nodes, extract the latitude and longitude coordinates and altitude coordinates of these key nodes, and organize them in the format of node number-longitude-latitude-altitude to form a set of key node spatial coordinates; then, for all nodes in the coordinate set... Point coordinates are sorted by ordinate (latitude), and the node with the smallest latitude value is selected as the reference point for the convex hull algorithm calculation. Then, the Graham scan method is used to execute the convex hull algorithm, sorting all key node coordinates clockwise according to the polar angle with the reference point, and judging the direction of adjacent three points in turn, i.e., left turn, right turn, or collinearity. Concave points are removed and convex points are retained, finally obtaining an ordered sequence of boundary points that constitute the minimum convex boundary. The sequence contains the boundary point numbers and their corresponding spatial coordinates. Finally, according to the order of the boundary point sequence, Geographic Information System (GIS) tools are used to connect each boundary point on an electronic map to form the minimum circumscribed convex polygon that can completely enclose all key nodes. The geographic area corresponding to this convex polygon is determined as the polygon monitoring area corresponding to this spatial domain calibration, ensuring that the monitoring area covers all key monitoring nodes and does not include irrelevant external spatial areas.

[0039] Step 133: Divide the polygonal monitoring area into regular grids, calculate the ratio of the area of ​​each grid cell to the total area of ​​the polygon, and obtain the area weight coefficient of each grid. Specifically, this includes: First, determining the grid size based on the actual area of ​​the polygonal monitoring area and the distribution density of key nodes. If the monitoring area is small (e.g., less than 10,000 square meters), a 10m × 10m grid cell is used; if the area is large (e.g., greater than 10,000 square meters), a 20m × 20m grid cell is used. This ensures that the number of grids after division meets the spatial refinement requirements without increasing the computational burden due to an excessive number of grids. Then, using GIS tools, the polygon monitoring area is divided into regular grids, generating multiple rectangular grid cells of the same size. The unique identifier number, coordinates of the four vertices, and coordinates of the grid center point are recorded for each grid cell. Next, the actual area of ​​each grid cell is calculated using the area calculation function of the GIS system. If a grid cell extends beyond the polygon monitoring area, only the area within the area is calculated, and the total area of ​​the entire polygon monitoring area is also calculated. Finally, the area of ​​each grid cell is divided by the total area of ​​the polygon monitoring area to obtain the area ratio of each grid cell. This ratio is the area weight coefficient of each grid cell.

[0040] Step 134: Based on the area weighting coefficients of each grid and combined with the distribution characteristics of pre-stored historical risk data within the grid, a dynamic adjustment coefficient is calculated using a weighted average algorithm. Specifically, this includes: first, calling the system's pre-stored historical risk database of the gas pipeline network to extract historical risk data corresponding to the polygonal monitoring area. This data includes records of gas leaks, pipeline corrosion, and third-party construction damage incidents that occurred in the area over the past 5 to 10 years. Each record contains the specific grid unit identifier of the incident, the incident type, the incident severity, and the incident time. The incident severity is categorized into three levels based on the scope of impact: minor, moderate, and severe. Then, statistical analysis is performed on the historical risk data, categorized by grid unit identifier... The system calculates the accident frequency and severity weighted value for each grid cell. The accident frequency is the number of accidents per unit time, and the severity weighted value is 1 point for minor accidents, 3 points for moderate accidents, and 5 points for severe accidents. The frequency and weighted value are summed according to the number of accidents, and the combination of frequency and weighted value forms the historical risk characteristic value of each grid. The larger the characteristic value, the higher the historical risk of the grid. Then, the area weight coefficient of each grid is multiplied by the corresponding historical risk characteristic value to obtain the weighted risk value of the grid. Finally, the weighted risk values ​​of all grids are summed and divided by the total number of grids to obtain the dynamic adjustment coefficient of the entire polygon monitoring area. This coefficient integrates the influence of spatial area weight and historical risk characteristics.

[0041] Step 135: Multiply the initial pipeline safety status assessment index by the dynamic adjustment coefficient to complete the spatial domain calibration process and obtain the comprehensive pipeline safety status assessment index. Specifically, this includes: First, extracting the initial pipeline safety status assessment index; determining the grid cell to which each sensor node belongs based on the sensor node number corresponding to the index (by matching the node coordinates with the grid cell coordinate range); averaging the initial assessment indices of all sensor nodes within the same grid cell to obtain the average initial assessment index for each grid cell; then extracting the dynamic adjustment coefficient corresponding to each grid cell (if multiple grids share the same adjustment coefficient, the unified coefficient is directly used); multiplying the average initial assessment index of each grid cell by the corresponding dynamic adjustment coefficient to complete the spatial domain calibration of that grid cell; finally, organizing all the calibrated assessment indices of all grid cells according to the grid identification number to form a set of assessment indices covering the entire polygonal monitoring area. This set is the comprehensive pipeline safety status assessment index. The calculation process for each index is also recorded (including the source of the initial index, grid affiliation determination, and the basis for calculating the adjustment coefficient).

[0042] In this embodiment of the invention, the initial pipeline safety status assessment index is received, and the optimal sensor node topology constituting the distributed monitoring network is simultaneously acquired. This allows for understanding the actual spatial distribution patterns of the sensor nodes, providing a precise network structure basis for subsequent delineation of monitoring areas and calculation of spatial weights. Extracting the spatial coordinates of key sensor nodes to form a coordinate set transforms the abstract topology into concrete spatial location data. Using the point with the smallest ordinate as the reference point, the convex hull algorithm is executed, automatically determining the minimum convex boundary based on the actual node distribution. A minimum circumscribed convex polygon is formed according to an ordered sequence of boundary points as the monitoring area, which not only fully covers the monitoring range of all key sensor nodes but also excludes irrelevant external spaces. Regular grid division of the polygon monitoring area transforms continuous spatial regions into discrete, quantifiable units. The area of ​​each grid unit is calculated. The area weight coefficient is obtained by proportionally relating the area to the total area of ​​the polygon, ensuring that the weight of different grids matches their actual spatial proportion. Based on the area weight coefficient of each grid, and combined with the distribution characteristics of historical risk data within the grid, the weight of the spatial dimension can be combined with the information of the historical risk dimension. The dynamic adjustment coefficient is calculated by a weighted average algorithm, which allows the adjustment coefficient to reflect both the spatial importance of the grid and the historical risk level. The initial pipeline safety status assessment index is multiplied by the dynamic adjustment coefficient to complete the spatial domain calibration, which can directly integrate the spatial weight and historical risk information into the initial index, thereby optimizing the spatial dimension of the initial index. The final comprehensive pipeline safety status assessment index not only includes the original safety status information, but also integrates the influence of spatial distribution differences and historical risk characteristics, which can more comprehensively and accurately reflect the actual safety status of different spatial areas.

[0043] In the underground gas pipeline safety early warning system described in this embodiment of the invention, step 132 includes: Step 1321: Extract the spatial coordinates of all key sensor nodes in the optimal sensor node topology to form a set of key node spatial coordinates. This includes: First, calling the complete database of the optimal sensor node topology, which contains the device attribute information of all sensor nodes, where the device attributes are marked as key monitoring nodes, backup redundant nodes, and communication relay nodes, etc.; Based on the safety monitoring requirements of underground gas pipeline networks, screen out the devices marked as key monitoring nodes. These nodes are equipped with micro-displacement sensors, methane concentration sensors, or mechanical vibration sensors, which are directly used to collect data on pipeline mechanical deformation, gas leakage, and external disturbances; Then, extract the spatial coordinate information of these key monitoring nodes from the database. The coordinate information includes latitude and longitude (accurate to 6 decimal places) and underground burial depth (accurate to centimeters), and organize them one by one in the format of node unique identifier number-longitude value-latitude value-burial depth value-sensor type; Finally, integrate the coordinates and related information of all key nodes to form a set of key node spatial coordinates.

[0044] Step 1322: Identify the point with the smallest ordinate in the key node spatial coordinate set as the reference point. If multiple points have the same ordinate, select the point with the smallest abscissa. Specifically, this involves: first, extracting the latitude values ​​(i.e., ordinates) of all nodes from the key node spatial coordinate set; sorting the latitude values ​​in ascending order using a data sorting algorithm; and selecting the node with the smallest latitude value. If, after sorting, two or more nodes have the same latitude value, further extracting the longitude values ​​(i.e., abscissas) of these nodes; sorting the longitude values ​​in ascending order; and selecting the node with the smallest longitude value as the reference point. Then, record the complete information of the reference point, including the node's unique identifier, longitude value, latitude value, burial depth, and the type of sensor mounted on it. Finally, mark the reference point information as the starting reference point for the convex hull algorithm calculation.

[0045] Step 1323: Using the reference point as the pole, calculate the polar angles of all other points in the key node spatial coordinate set relative to the pole, and sort all points in ascending order of polar angle to form an ordered point set. Specifically, this includes: First, using the reference point determined in step 1322 as the pole, and the due north direction of the reference point's location as the polar axis (0-degree direction), calculate the polar angle of each other node in the key node spatial coordinate set relative to the reference point using the geographic coordinate angle calculation method; When calculating the polar angle, based on the longitude and latitude difference between the node and the reference point, convert it into a clockwise angle centered on the reference point (range from 0 degrees to 360 degrees) through trigonometric functions; If there are two or more nodes with the same polar angle relative to the reference point, further calculate the straight-line distance between these nodes and the reference point (based on the surface distance formula of longitude and latitude), and sort them a second time in order of distance from the reference point from near to far; Then, arrange all nodes (including the reference point) in ascending order of polar angle, and arrange those with the same polar angle in order of distance from near to far to form an ordered point set.

[0046] Step 1324 involves traversing the ordered point set using a stack data structure, determining the turning relationship of the points based on the cross product of continuous vectors, removing intermediate points that do not satisfy the convex hull condition, and obtaining an ordered boundary point sequence that constitutes the minimum convex boundary. Specifically, this includes: first, initializing a stack data structure and pushing the first two points (the reference point and the point with the smallest polar angle) from the ordered point set formed in step 1323 onto the stack; then, starting from the third point in the ordered point set, traversing all remaining points one by one, taking the top two points of the stack (denoted as points A and B) and the currently traversed point (denoted as point C) in each traversal; calculating the cross product of vector AB (pointing from A to B) and vector AC (pointing from A to C), and determining the turning relationship of the three points based on the cross product result. If the cross product result is positive, it indicates that the three points are in a left-turning state. Point C satisfies the convex hull boundary condition, so it is pushed onto the stack. If the cross product is negative, it means the three points are in a right-turn state, and point B does not satisfy the convex hull boundary condition, so point B is popped from the stack. The process continues, taking the top two points of the stack and calculating the cross product with the current point C, until the cross product is positive or only one point remains in the stack. Then, point C is pushed onto the stack. If the cross product is zero, it means the three points are collinear. The distances from points B and C to point A are calculated, keeping the points with the greater distance and popping the points with the smaller distance from the stack. After traversal, the points stored in the stack are the points that constitute the minimum convex boundary. These points are extracted according to the order in which they are pushed onto the stack, forming an ordered sequence of boundary points. Finally, the unique identifier and coordinates of each point in the ordered sequence of boundary points are recorded to ensure that the sequence order is consistent with the clockwise or counterclockwise direction of the convex hull boundary.

[0047] Step 1325: Connect the boundary points sequentially according to the ordered boundary point sequence of the minimum convex boundary to form the minimum circumscribed convex polygon, and determine the polygon as the corresponding polygon monitoring area. Specifically, this includes: First, importing the ordered boundary point sequence into a Geographic Information System (GIS) software, loading an electronic map (containing basic geographic information such as surface topography, buildings, and other underground pipelines) of the area where the underground gas pipeline network is located in the software; drawing line segments connecting each boundary point on the electronic map according to the order of the ordered boundary point sequence, ensuring that the line segments precisely match the latitude and longitude coordinates of each point; finally, connecting the last point of the sequence with the first point to form a closed polygon; then verifying in the GIS software whether the polygon completely encloses all the key sensor nodes in step 1321. If any key nodes are not enclosed, checking whether the ordered boundary point sequence is missing, re-executing the cross product judgment process in step 1324, and supplementing the missing boundary points; after successful verification, extracting the spatial parameters such as the boundary coordinate range and total area of ​​the polygon in the GIS software, labeling the key node numbers and sensor types contained within the polygon, and finally determining the closed polygon as the corresponding polygon monitoring area.

[0048] In this embodiment of the invention, the spatial coordinates of all key sensing nodes in the optimal sensing node topology are extracted and formed into a set. This allows for focusing on the data of core monitoring nodes and eliminating interference from non-critical nodes such as backup nodes and redundant communication nodes. It ensures that subsequent convex hull algorithm calculations are based solely on the coordinates of nodes directly involved in pipeline safety monitoring, avoiding boundary calculation deviations caused by irrelevant coordinate data. The point with the smallest ordinate in the key node spatial coordinate set is identified as the reference point. If multiple points have the same ordinate, the point with the smallest abscissa is selected, allowing for the determination of a unique reference point through explicit rules. The polar angles of the remaining points are calculated using the reference point as the pole, and the points are sorted in ascending order of polar angles to form an ordered point set. This organizes the scattered node coordinates according to a circular order around the reference point. A stack data structure is used to traverse the ordered point set, relying on the connection... The vector cross product is used to determine the turning relationship of points and remove intermediate points that do not meet the convex hull condition. The vector cross product can accurately distinguish whether a point is in a valid left-turn state or an invalid right-turn state at the convex hull boundary, thereby efficiently eliminating non-boundary points such as concave points and collinear redundant points, ensuring that the final ordered boundary point sequence can accurately form the minimum convex boundary, and improving the accuracy and effectiveness of boundary point selection. The points are connected sequentially according to the ordered boundary point sequence of the minimum convex boundary to form the minimum circumscribed convex polygon, which is determined as the monitoring area, which can completely cover the monitoring range of all key sensing nodes. This ensures that the monitoring area is accurately matched with the actual monitoring capability, ensuring that all key monitoring points are within the monitoring area, without increasing the redundant workload of subsequent mesh division and weight calculation due to the area being too large, providing a reasonable and compact spatial analysis range for subsequent spatial domain calibration.

[0049] In the underground gas pipeline safety early warning system described in this embodiment of the invention, step 134 includes: Step 1341 involves receiving the area weight coefficients of each grid and querying the pre-stored historical risk database to obtain the historical risk level data corresponding to each grid. Specifically, this includes: first, receiving the area weight coefficients of each grid through the system's internal data interaction interface. These coefficients are transmitted as key-value pairs of the grid's unique identifier and its area weight value. After receiving the data, they are stored in a dedicated weight data cache module. Simultaneously, the system verifies whether the area weight value of each grid is within a reasonable range of 0 to 1. If any abnormal value exceeds this range, a data retransmission mechanism is triggered to ensure the accuracy of the received area weight coefficients. Then, the system's pre-stored historical risk database is invoked. This database stores the risk levels of various areas of the underground gas pipeline network over the past 5 to 10 years. The accident record includes the specific grid's unique identifier number, accident type, accident time, accident impact range, accident handling results, and risk level assessment results. Accident types include soil corrosion leaks, third-party construction damage, and pipeline aging and rupture. The risk level assessment results are determined by gas safety experts based on the degree of accident loss and impact range. Based on the unique identifier number of each grid, a precise matching query is performed in the historical risk database to extract the risk level assessment results corresponding to all accident records of each grid within the past 5 to 10 years. If a grid has no historical accident records, it is marked as a risk-free record, ultimately forming a correspondence table between the grid's unique identifier number and historical risk level data.

[0050] Step 1342: Based on the historical risk level data, the risk level of each grid is quantified into a corresponding risk weight coefficient through a risk level-weight mapping table. Specifically, this includes: first, retrieving the pre-configured risk level-weight mapping table from the system. This mapping table is determined by combining the risk characteristics of the underground gas pipeline network with expert review in the field of gas safety and statistical analysis of historical accidents. The table clearly classifies risk levels into four levels: high risk, medium risk, low risk, and extremely low risk. High risk corresponds to two or more serious leaks or construction damage accidents within the past three years, affecting three or more adjacent grids. Medium risk corresponds to one serious accident or two or more minor leaks within the past three years, affecting only its own grid. Low risk corresponds to only one minor leak accident within the past five years with no significant property damage. Extremely low risk corresponds to no accident records within the past five years. The risk level may have only experienced minor deformations that can be repaired immediately, and each risk level corresponds to a fixed range of risk weight coefficients; high risk corresponds to 0.8 to 1.0, medium risk to 0.5 to 0.7, low risk to 0.2 to 0.4, and very low risk to 0.05 to 0.1. Then, based on the correspondence table between the grid's unique identifier number and historical risk level data, the risk level-weight mapping table is matched one by one according to the grid's unique identifier number. If the grid's historical risk level is high, 0.9 (taking the middle value of the range to ensure stable weight values) is selected as the risk weight coefficient for that grid; 0.6 is selected for medium risk, 0.3 for low risk, and 0.08 for very low risk; if the grid is marked as a risk-free record, a risk weight coefficient of 0.08 is assigned according to the very low risk standard. Finally, the unique identifier number of each grid is associated with and stored with the matched risk weight coefficients to form a correspondence table between the grid's unique identifier number and risk weight coefficients.

[0051] Step 1343 involves weighting and summing the area weight coefficients of each grid with the corresponding risk weight coefficients to generate a dynamic adjustment coefficient. This process includes: first, matching each grid's unique identifier with its corresponding area weight value in the grid unique identifier-risk weight coefficient correspondence table to ensure accurate matching and avoid calculation errors caused by identifier confusion; then, for each successfully matched grid, performing a weighted calculation by multiplying its area weight value by the corresponding risk weight coefficient to obtain its comprehensive weight contribution value. For example, if a grid has an area weight value of 0.05 and a risk weight coefficient of 0.9, its comprehensive weight contribution value is 0.05 × 0.9 = 0.045; after calculating the comprehensive weight contribution values ​​for all grids, summing these values ​​yields the total weight contribution value for the entire polygon monitoring area. Finally, this total weight contribution value is used as the dynamic adjustment coefficient.

[0052] In this embodiment of the invention, receiving the area weight coefficients of each grid clarifies the spatial importance of different grids. Simultaneously, querying a pre-stored historical risk database retrieves historical risk level data for each grid, incorporating past risk occurrence characteristics and linking spatial area information with historical risk information over time. Based on the historical risk level data, a risk level-weight mapping table quantifies the risk level into risk weight coefficients, transforming qualitative risk levels (such as high, medium, and low risk) into quantitative numerical weights. This solves the problem that qualitative data cannot directly participate in mathematical calculations, enabling historical risk characteristics to be integrated into adjustment coefficient calculations using a unified quantitative standard. Weighted summation of the area weight coefficients of each grid with their corresponding risk weight coefficients generates a dynamic adjustment coefficient, merging the objective proportion of spatial area with the characteristic weights of historical risks into a single adjustment parameter. This ensures that the dynamic adjustment coefficient reflects both the spatial importance of the grid within the monitoring area and the past probability of risk occurrence, avoiding the bias caused by adjustment coefficients being determined solely by spatial or historical factors, and making the generated dynamic adjustment coefficients more closely aligned with the actual risk attributes of each grid.

[0053] In the underground gas pipeline safety early warning system described in this embodiment of the invention, module 14 receives comprehensive pipeline safety status assessment indicators, transmits them through a dual-mode communication link, and uses edge computing nodes to relay and buffer the transmitted data, successfully uploading the indicators to the monitoring center to obtain pipeline safety status data located at the monitoring center, including: Step 141: Receive the comprehensive pipeline safety status assessment index and encapsulate it into a data packet conforming to the dual-mode communication protocol. Specifically, this includes: firstly, conducting a transmission characteristic analysis of LoRa and NB-IoT communication links to clarify the bandwidth limitations, frame structure requirements, and data format specifications of the two links. The bandwidth limitations are: LoRa commonly uses 125kHz / 250kHz bandwidth, and NB-IoT commonly uses 180kHz bandwidth. The frame structure requirements are: LoRa must include a preamble, synchronization word, and physical layer frame header; NB-IoT must conform to the NB-IoT physical layer protocol frame structure defined by 3GPP. The data format specifications are: both support binary and hexadecimal data transmission, and special characters should be avoided to prevent parsing errors. This ensures that the template design is compatible with the hardware transmission capabilities and protocol standards of both links.

[0054] Next, the core fields of the data packet were determined. Based on the transmission requirements of the comprehensive pipeline safety status assessment indicators, the following fields were selected as essential: protocol version number, data type identifier, grid unique identifier number, assessment value, acquisition timestamp, checksum, and end-of-transmission identifier. The protocol version number is used for version compatibility identification during subsequent template upgrades to avoid data parsing conflicts between old and new systems. The data type identifier clearly distinguishes pipeline safety assessment data from other types of data within the system, such as raw sensor data, facilitating rapid classification and processing at the receiving end. The grid unique identifier number accurately locates the monitoring grid corresponding to the assessment data, supporting subsequent spatial domain risk tracing. The assessment value is the core data, reflecting the grid's safety status. The acquisition timestamp records the data generation time for time-series risk analysis. The checksum detects data corruption or loss during transmission. The end-of-transmission identifier clearly defines the data packet boundaries, preventing data concatenation and parsing errors at the receiving end.

[0055] Then, the format and length of each field are defined: the protocol version number is set to 1 byte, binary format, supporting versions 0 to 255 to meet long-term upgrade needs; the data type identifier is set to 2 bytes, hexadecimal format, with pipeline safety assessment data corresponding to a fixed identifier of 0x0102 for easy and quick identification; the grid unique identifier number is set to 8 bytes, decimal to hexadecimal format, covering tens of millions of grids to meet the monitoring needs of large-scale urban gas pipeline networks; the assessment value is set to 4 bytes, floating-point data to hexadecimal format, retaining 4 decimal places of precision to ensure the accuracy of the assessment results; the acquisition timestamp is set to 8 bytes, UTC time to hexadecimal format, accurate to the millisecond level, supporting time-series data sorting; the checksum is set to 2 bytes, using the CRC16 algorithm, with a calculation range covering all fields from the protocol version number to the acquisition timestamp, effectively detecting single-bit, double-bit errors and parity errors; the end identifier is set to 1 byte, fixed value 0xFF, clearly marking the end of the data packet.

[0056] Next, field format conversion rules were established, with a unified conversion logic preset for the original data types of different fields: evaluation values ​​were converted from decimal floating-point to hexadecimal, such as decimal 0.85 to hexadecimal 0x3F583126; collection timestamps were converted from UTC time strings to hexadecimal, such as 2024-05-2014:30:00.123 to the corresponding hexadecimal timestamp; other fields were converted according to preset fixed formats to ensure that the conversion logic of the sending end and the receiving end is consistent and to avoid parsing deviations.

[0057] Finally, the above field structure, length, format conversion rules, and checksum algorithm are integrated into a complete dual-mode communication protocol encapsulation template, which is stored in the firmware and software configuration file of the system communication module. At the same time, the template compatibility is verified through actual hardware testing. The encapsulated data packets are sent using a LoRa module (operating frequency band 470 to 510MHz) and an NB-IoT module (accessing the operator's NB-IoT network). The integrity and accuracy of data parsing are verified at the receiving end. Fine-tuning and optimization are carried out for field length adaptation issues and format conversion errors that occur during the test. Finally, the template preset is completed to ensure that data packets that meet the transmission requirements of both links can be directly generated when called in subsequent times.

[0058] Step 142: Send the data packet through the primary LoRa communication link and monitor the link signal quality in real time to generate link status monitoring results. Specifically, this includes: first, starting the primary LoRa communication module configured in the system and setting the module's operating parameters, including the operating frequency band, spreading factor, bandwidth, and transmission power. The operating frequency band is 470 to 510 MHz, conforming to domestic industrial IoT communication standards; the spreading factor is SF=12, suitable for underground environment signal transmission requirements; the bandwidth is 125 kHz; and the transmission power is 20 dBm. Then, send the encapsulated data packet through the LoRa module's RF interface, and simultaneously activate the module's built-in signal quality monitoring function to collect link status data in real time. The two key parameters of the link, Received Signal Strength Indication (RSSI) and Signal-to-Noise Ratio (SNR), are collected every 100ms. The RSSI value (unit: dBm) and SNR value (unit: dB) of each collection are recorded to the link status log and stored in the format of collection time-data packet number-RSSI value-SNR value. At the same time, the current parameters are compared with the system's preset normal thresholds in real time. RSSI is greater than or equal to -110dBm and SNR is greater than or equal to 5dB. If the parameters collected for 3 consecutive times are within the normal threshold range, the link status is marked as stable. If the parameters are lower than the threshold, the link status is marked as unstable. Finally, a real-time updated link status monitoring result is generated.

[0059] Step 143: Based on the link status monitoring results, a judgment is made. When the signal quality is determined to be lower than a preset threshold, a link switching command is executed to establish a valid connection with the backup NB-IoT communication link. Specifically, this includes: first, reading the link status monitoring results and judging the status flag of the primary LoRa link. If the flag is unstable and the duration of this state reaches 3 seconds (3 seconds is the system's preset switching trigger duration), then it is determined that the current primary link signal quality is lower than the preset threshold, triggering the link switching process; first, a shutdown command is sent to the primary LoRa communication module to stop its data transmission and signal monitoring functions, and then the backup NB-IoT communication module is started. The module automatically searches for NB-IoT network signals from surrounding operators, completes network registration, and obtains a dynamic IP address; a connection request is sent through the module's built-in AT command set to establish a reliable TCP connection with the system's preset NB-IoT communication server. After successful connection, the server returns a connection confirmation command; after receiving the confirmation command, a signal indicating that the backup link has been established is sent back to the system, completing the valid connection with the backup NB-IoT communication link and ensuring that the data transmission link is not interrupted.

[0060] Step 144: Through the established effective NB-IoT communication link, the data packet is transmitted to the edge computing node along the pipeline. This node performs receiving, relaying, and buffering operations on the data packet to obtain a stable data transmission stream. Specifically, this includes: firstly, through the established effective NB-IoT communication link, sending the data packet encapsulated in Step 141 to the edge computing node along the pipeline at a frequency of one packet every 500ms. The edge computing nodes are deployed at intervals of one every 2 kilometers, close to the inspection wells of the gas pipeline. The node hardware configuration includes an ARM Cortex-A53 processor, a 4G / 5G communication module, and a 128GB SSD storage unit. After receiving the data packet, the edge computing node first verifies the protocol version number and data type identifier in the data packet header to confirm... After the pipeline safety assessment data is received, the built-in CRC check algorithm is used to verify the checksum and determine whether the data packet is complete. For complete data packets that pass the verification, the node performs a relay operation. If the communication distance between the node and the monitoring center is within 5 kilometers, the node directly forwards the data packet to the monitoring center through its own 4G / 5G module. If the distance exceeds 5 kilometers, the data packet is forwarded to the next adjacent edge computing node for relay transmission. At the same time, the node stores each complete data packet in the local SSD storage unit, classifying and saving it according to the format of reception time-link type-data packet content, with a storage period of 24 hours, which facilitates subsequent data traceability. Through the coordinated operation of relay and caching, a continuous and lossless stable data transmission stream is formed, avoiding data interruption due to signal fluctuations during long-distance transmission.

[0061] Step 145: Perform integrity verification on the data packets that successfully arrive at the monitoring center in the transmitted data stream, and filter out the valid data packets that pass the verification; perform protocol parsing on the valid data packets to extract the payload data; reassemble the payload data according to a predefined data structure to restore the pipeline safety status data. Specifically, this includes: First, the receiving server at the monitoring center splits the received transmitted data stream into data packets, extracting the successfully arrived data packets one by one according to the data packet number; performing integrity verification on each data packet, calling the CRC16 check algorithm consistent with step 141, calculating the check value of the data packet content and comparing it with the check code carried in the packet. If the two match, it is determined to be a valid data packet that has passed the verification; if they do not match, it is marked as a corrupted data packet and discarded; perform protocol parsing on the valid data packets. Following the structure of the dual-mode communication protocol encapsulation template, payload data such as protocol version number, grid unique identifier number, evaluation value, and collection timestamp are extracted sequentially from the data packets, removing redundant fields such as header protocol identifier and tail end identifier. The extracted payload data is then reassembled according to a predefined data structure, which is in tabular form and includes five columns: data reassembly time, grid ID, pipeline safety assessment value, data collection time, and transmission link type (NB-IoT). The payload data of each valid data packet is filled into the corresponding fields. After reassembly, the tabular data is written to the MySQL database of the monitoring center, and a data reassembly log is generated, recording the reassembly time, the number of data packets processed, and the number of valid data packets. Finally, pipeline safety status data that can be directly used for subsequent early warning analysis is restored.

[0062] In this embodiment of the invention, the comprehensive pipeline safety status assessment index is encapsulated into a data packet conforming to the dual-mode communication protocol, enabling the assessment index data to adapt to the transmission format requirements of both LoRa and NB-IoT communication links. Sending data packets via the primary LoRa communication link leverages the wide coverage and low power consumption advantages of the LoRa link in underground environments to achieve initial efficient data transmission. Real-time monitoring of link signal quality and generation of monitoring results allows for timely understanding of the primary link's transmission status, such as signal strength and packet loss rate. Based on the link status monitoring results, when the signal quality falls below a preset threshold, a link switching command is executed and a backup NB-IoT link connection is established. The redundancy design of the dual-mode link avoids data transmission interruptions caused by a single link failure. The backup link quickly replaces the poorly functioning primary link, ensuring the continuity of data transmission and reducing transmission delays caused by link problems. Effective NB-IoT... The IoT communication link transmits data packets to edge computing nodes along the pipeline. These edge computing nodes perform receiving, relaying, and caching operations, temporarily storing and stabilizing data during transmission to the monitoring center. This avoids data loss or delay due to signal fluctuations during long-distance direct transmission to the monitoring center, while also reducing the pressure on the monitoring center to directly receive large amounts of scattered data, forming a stable data transmission stream and providing intermediate assurance for the data's safe arrival at the monitoring center. The link also performs integrity verification on data packets arriving at the monitoring center, ensuring data quality for subsequent processing. Furthermore, it performs protocol parsing on valid data packets and extracts payload data, restoring the encapsulated data packets to recognizable original data fragments. Finally, it reassembles the payload data according to a predefined data structure, integrating scattered data fragments into complete and ordered pipeline safety status data, ensuring that the monitoring center obtains accurate, complete, and directly usable data for risk analysis.

[0063] In the underground gas pipeline safety early warning system described in this embodiment of the invention, the above-mentioned module 15 performs dynamic risk analysis on the pipeline safety status data to obtain analysis results. When the analysis results exceed a preset threshold, an early warning is automatically triggered, and linkage instructions are simultaneously sent to the gas operation and maintenance platform, construction unit, and community emergency system to achieve closed-loop management, including: Step 151: Input the pipeline safety status data into the risk quantification scoring model for dynamic analysis to obtain the multi-dimensional analysis results of the risk quantification scoring model. Specifically, this includes: First, receiving pipeline safety status data through the data interface of the monitoring center. This data includes the unique identifier number of each grid, the comprehensive pipeline safety assessment value, the data collection timestamp, the latitude and longitude range of the corresponding monitoring area, the transmission link type, and the data calculation basis. After receiving the data, the integrity of the data is first checked. After confirming that there are no missing fields (such as grid ID, assessment value) or abnormal values ​​(such as negative assessment values), the data is temporarily stored in the risk analysis database according to the grouping method of grid ID-collection time. Then, the risk quantification scoring model deployed in the system is invoked. The model construction process is as follows: First, considering the multi-dimensional requirements of underground gas pipeline network risk assessment to cover real-time status, historical characteristics, spatial attributes, and environmental impact, the model input dimensions are determined as real-time comprehensive assessment value, grid historical risk frequency (number of accidents in the past 5 years), grid spatial weight coefficient (calculated in step 133), and surrounding environmental interference coefficient (such as whether there is a construction area and soil corrosion level). Then, preset data preprocessing rules are formulated for each input dimension, for example, the grid historical risk frequency is set as 0 for 0 times, 0.3 for 1 to 2 times, 0.3 for 3 times, and so on. The mapping relationship of 0.7 for 4th time and 1 for 5th time and above is converted into a normalized coefficient of 0 to 1. The surrounding environmental interference coefficient is quantified according to the standard of 0.1 for no construction and low soil corrosion level, 0.5 for temporary construction or medium soil corrosion level, and 0.9 for long-term construction and high soil corrosion level. Then, the weighted association algorithm is selected as the core calculation logic of the model. This algorithm can adapt to the linear fusion of multi-dimensional data. At the same time, based on the experience of gas safety experts, the weights of each dimension are initially set: real-time comprehensive evaluation value 0.4, historical risk frequency 0.25, spatial weight coefficient 0.2, and environmental interference coefficient 0.15. The training process of the model is as follows: First, historical operation data of underground gas pipeline networks over the past 10 years are collected to construct a training dataset. Each sample in the dataset contains historical data for the four input dimensions mentioned above and a label indicating whether an accident has occurred in the corresponding grid (1 for an accident, 0 for no accident). Then, the training dataset is divided into a training set and a validation set in a 7:3 ratio. The model is iteratively trained using the training set. During each iteration, the error between the model's output risk prediction value and the sample label is calculated. The weights of each dimension are adjusted using gradient descent to reduce the error. Every 100 rounds of training, the model's prediction accuracy is tested using the validation set to determine the match between the risk level and the actual accident. Training stops when the accuracy is consistently above 90% for three consecutive rounds, and the final weights of each dimension are determined, i.e., the weights optimized based on historical accident data. Finally, the model's performance is validated using new operation data from the past year to ensure the model's analytical stability under different scenarios (such as high soil humidity during the rainy season and periods of intensive construction), thus completing the model training. The implementation process of the model is as follows: The trained model is encapsulated into a modular program that can be called by the system, integrating a data preprocessing module and a weighted correlation calculation module. The preprocessing module automatically performs data normalization and quantization for each dimension, and the weighted correlation calculation module performs multi-dimensional data fusion according to the final weight. The model is deployed to the edge computing node of the monitoring center to ensure the response speed when the model is called, with a single analysis taking no more than 1 second. At the same time, a data interface is reserved to support the model expansion when adding new input dimensions (such as pipeline material aging coefficient). Next, the temporarily stored pipeline safety status data is matched and preprocessed according to the model input dimensions. For example, real-time comprehensive assessment values ​​are extracted from the data. No additional conversion is required, as they are already dimensionless values ​​from 0 to 1. The accident counts for the corresponding grid in the past 5 years are obtained from the historical risk database and converted into normalized coefficients according to the preprocessing rules. The grid spatial weight coefficients are retrieved and directly adapted to the model input format. The current construction status and soil corrosion level of the grid are obtained through the environmental monitoring subsystem and quantified into the surrounding environmental interference coefficient according to the rules. Finally, the preprocessed data of each dimension is input into the risk quantification scoring model. The model performs dynamic analysis on the data through a weighted correlation algorithm (using the final weights determined by training), and calculates the contribution of each dimension of data to pipeline risk in real time. For example, when the real-time comprehensive assessment value is 0.8, its contribution is calculated to be 0.32 with a weight of 0.4. The final output includes a multi-dimensional analysis result containing grid ID, risk contribution value of each dimension, and risk correlation (co-influence coefficient of contribution value of each dimension).

[0064] Step 152: Based on the multi-dimensional analysis results of the risk quantification scoring model, calculate the risk score value of the current pipeline safety status. Specifically, this includes: first, extracting the multi-dimensional analysis results of each grid, determining the contribution of real-time assessment values, historical risk contributions, spatial weight contributions, and environmental interference contributions for each grid; then, retrieving the system's preset dimension weight allocation table, which was determined through gas safety expert review and regression analysis of gas pipeline accident data from the past 10 years. The table assigns a weight of 40% to real-time assessment value contributions, 25% to historical risk contributions, and 25% to spatial weight contributions. The risk contribution of each dimension is weighted at 20%, and the contribution of environmental disturbance is weighted at 15%, with the total weight of all dimensions being 1. For each grid, the risk contribution value of each dimension is multiplied by its corresponding weight to obtain the weighted contribution value of each dimension. For example, if the real-time assessment contribution of a grid is 0.8 and the corresponding weight is 0.4, then the weighted contribution value is 0.8 × 0.4 = 0.32. The weighted contribution values ​​of all dimensions are added together to obtain the basic risk score of the current pipeline safety status of the grid. The basic score is then converted into a risk score value from 0 to 100 according to the conversion rule of basic score × 100. The higher the score, the higher the risk.

[0065] Step 153: Compare the risk score with multiple preset risk thresholds to determine the corresponding risk level. Specifically, this includes: first, retrieving multiple pre-stored risk threshold standards. These standards are formulated based on the national "Urban Gas Management Regulations" and local gas pipeline safety operation specifications, while also referencing the risk scoring characteristics of major gas accidents in the past five years. Four risk threshold ranges are preset: low risk, medium risk, high risk, and extremely high risk. Low risk is 0 to 30 points, representing a stable pipeline with no obvious safety hazards. Medium risk is 31 to 60 points, representing a minor pipeline abnormality requiring close monitoring. High risk is 61 to 85 points, representing a pipeline with... In cases of significant risk, inspections and investigations must be initiated. Extremely high risk is defined as a score of 86 to 100, indicating an emergency pipeline requiring immediate action. Then, the risk score for each grid is compared with a preset risk threshold range, grid ID by grid. For example, a grid with a risk score of 72 is classified as high-risk. If a grid's score falls within the range boundary (e.g., 30 or 60), its historical risk trend (score change over the past 7 days) is considered for further confirmation. For instance, a score of 30 with no upward trend over the past 7 days is classified as low-risk; an upward trend indicates medium-risk. Finally, a unique risk level is assigned to each grid.

[0066] Step 154: When the risk level reaches the warning threshold, automatically generate multi-level warning information and response plans matching the risk level. Specifically, this includes: first, filtering out grids with high or extremely high risk levels; the system presets these two levels as warning thresholds, while low and medium risks do not trigger warnings; then, calling the system's multi-level warning information template library, which is categorized by risk level. High-risk warning information templates include the warning grid ID, risk score, risk location (latitude and longitude converted to a specific address), potential risk type (such as third-party construction interference, corrosion leakage hazards), and suggested response timeframe (within 24 hours); extremely high-risk warning information... The template adds urgency level, potential impact range, and temporary protective measures to the high-risk level. It also accesses a response plan database. High-risk areas correspond to routine inspection plans, specifying the number of inspectors, equipment such as methane detectors, and inspection routes. Extremely high-risk areas correspond to emergency repair plans, specifying the repair team composition, required equipment such as excavators and sealing tools, and repair procedures such as gas cut-off-detection-repair-restore. Based on the selected grid risk level, the template matches the corresponding early warning information and response plan, filling in specific information such as grid ID, address, and score, generating personalized multi-level early warning information and response plans for each high / extremely high-risk grid.

[0067] Step 155: Simultaneously push the multi-level early warning information and response plan to the gas operation and maintenance platform, construction unit operation terminals, and community emergency system, and receive response feedback information from each system to complete closed-loop management. Specifically, this includes: First, simultaneously pushing the multi-level early warning information and response plan to the corresponding recipients: Pushing complete early warning information and response plans to the gas operation and maintenance platform, including the early warning grid location marked on the electronic map and dispatch information for inspection / repair personnel; Pushing the early warning grid location and work stoppage notice to the construction unit operation terminals; if there is construction work around the early warning grid, confirming and requiring the suspension of construction operations that may affect the pipeline; Pushing the early warning impact range and resident notification content to the community emergency system; such as pushing evacuation notices in cases of extremely high risk, including the community emergency contact person and contact information; After the push is completed, start the feedback reception timer. The feedback time limit is 1 hour for high-risk situations and 15 minutes for extremely high-risk situations. The feedback interfaces of each receiving object are monitored in real time. The feedback content from the gas operation and maintenance platform must include the dispatch status (dispatched / not dispatched), the name of the personnel handling the situation, and the estimated departure time. The feedback content from the construction unit must include whether work has been stopped and the time of work stoppage confirmation. The feedback content from the community emergency system must include whether the information has been received and the emergency preparedness status (such as evacuation route confirmation). After receiving feedback information, the feedback time and content of each object are recorded. If no feedback is received within the time limit, a second push is automatically triggered (with an interval of 5 minutes) and marked as pending confirmation. After all receiving objects have provided feedback or the time limit has expired, the warning push time, receiving object, feedback content, and feedback time are compiled into a closed-loop management record table and stored in the emergency response archive to complete the closed-loop management from warning triggering to feedback tracking.

[0068] In this embodiment of the invention, pipeline safety status data is input into a risk quantification scoring model for dynamic analysis. The model's pre-set structured analysis logic (such as algorithms that integrate spatial weights, historical risks, and real-time monitoring data) allows for systematic decomposition and correlation of the data. Simultaneously, the dynamic analysis mode responds in real-time to changes in pipeline safety status data, adjusting the analysis dimensions synchronously with data updates, ensuring the model's multi-dimensional analysis results closely align with the pipeline's current actual safety status. Based on the multi-dimensional analysis results of the risk quantification scoring model, a risk score is calculated, converting the model's output multi-dimensional analysis information, such as spatial risk contribution, historical risk proportion, and real-time parameter anomaly levels, into quantifiable values. The risk score is compared with multiple pre-set risk thresholds to determine the risk level. Multiple thresholds allow for refined differentiation of risk levels, ensuring the risk level accurately reflects the actual severity of the risk corresponding to the score. Furthermore, the pre-set thresholds are based on gas pipeline network safety operation standards and historical data. Accident data calibration ensures that the risk classification results match the actual impact range and urgency of the safety risks. When the risk level reaches the warning threshold, multi-level warning information and response plans are automatically generated to match the risk level, achieving level adaptation between warning information and response measures. Different risk levels correspond to differentiated warning content and response steps, avoiding resource waste or insufficient response caused by using uniform content and plans for all warnings. Multi-level warning information and response plans are simultaneously pushed to the gas operation and maintenance platform, construction unit operation terminals, and community emergency systems, enabling real-time information synchronization among multiple entities and avoiding response gaps caused by information transmission delays or asymmetry. At the same time, response feedback information from each system (such as the departure time of operation and maintenance personnel, confirmation of work stoppage by construction units, and community emergency preparedness status) can be received, enabling real-time tracking of the response progress of each entity and connecting information push, response execution, and progress feedback into a complete process. Finally, closed-loop management is achieved to ensure that all relevant links can proceed in sequence after the warning is triggered.

[0069] like Figure 2 As shown, a control method for an underground gas pipeline network safety early warning system is disclosed, the control method comprising: Sensor nodes are deployed at key locations on the gas pipeline and its surrounding environment to form a distributed monitoring network; the mechanical deformation, methane concentration and mechanical vibration data of the pipeline are collected synchronously through the distributed monitoring network to obtain a multimodal sensing data set; Humidity and temperature interference corrections were applied to the methane concentration data in the multimodal sensing dataset to obtain corrected methane concentration data. The mechanical deformation data, mechanical vibration data and corrected methane concentration data were then fused together to obtain the initial pipeline safety status assessment index. The system receives the initial pipeline safety status assessment index, delineates the corresponding polygonal monitoring area based on the topology of the distributed monitoring network, and performs grid division to obtain the area weight coefficient of each grid. Based on this weight coefficient, a dynamic adjustment coefficient is obtained through weighted calculation, and the initial pipeline safety status assessment index is spatially calibrated using the dynamic adjustment coefficient to obtain a comprehensive pipeline safety status assessment index. The system receives comprehensive pipeline safety status assessment indicators, transmits them through a dual-mode communication link, and uses edge computing nodes to relay and cache the transmitted data. The indicators are then successfully uploaded to the monitoring center to obtain pipeline safety status data located at the monitoring center. Dynamic risk analysis is performed on pipeline safety status data to obtain analysis results. When the analysis results exceed the preset threshold, an early warning is automatically triggered and linkage instructions are sent to the gas operation and maintenance platform, construction unit and community emergency system simultaneously to achieve closed-loop management.

[0070] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the system as described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0071] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the system as described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A safety early warning system for underground gas pipeline networks, characterized in that, include: The data acquisition module is used to deploy sensor nodes at key locations on the gas pipeline itself and in the surrounding environment to form a distributed monitoring network. The mechanical deformation, methane concentration and mechanical vibration data of the pipeline are collected synchronously through a distributed monitoring network to obtain a multimodal sensing data set; The fusion module is used to correct for humidity and temperature interference in the methane concentration data in the multimodal sensing dataset to obtain corrected methane concentration data. The mechanical deformation data, mechanical vibration data and the corrected methane concentration data are fused together to obtain the initial pipeline safety status assessment index. The calibration module is used to receive the initial pipeline safety status assessment index, delineate the corresponding polygonal monitoring area based on the topology of the distributed monitoring network, and perform grid division to obtain the area weight coefficient of each grid. Based on this weighting coefficient, a dynamic adjustment coefficient is obtained through weighted calculation, and the initial pipeline safety status assessment index is spatially calibrated using the dynamic adjustment coefficient to obtain a comprehensive pipeline safety status assessment index. The transmission module is used to receive comprehensive pipeline safety status assessment indicators, transmit them through a dual-mode communication link, and use edge computing nodes to relay and cache the transmitted data, successfully uploading the indicators to the monitoring center to obtain pipeline safety status data located at the monitoring center. The early warning module is used to perform dynamic risk analysis on pipeline safety status data and obtain analysis results. When the analysis results exceed the preset threshold, an early warning is automatically triggered and linkage instructions are sent to the gas operation and maintenance platform, construction unit and community emergency system to achieve closed-loop management.

2. The underground gas pipeline safety early warning system according to claim 1, characterized in that, Sensor nodes are deployed at key locations within the gas pipeline and its surrounding environment to form a distributed monitoring network. By synchronously collecting data on pipeline mechanical deformation, methane concentration, and mechanical vibration through a distributed monitoring network, a multimodal sensing data set is obtained, including: Based on the laying path of the gas pipeline, the risk level of the surrounding environment and historical accident data, the optimal sensor node topology covering the pipeline body and adjacent key areas is obtained. Based on the topology, micro-displacement sensors, methane concentration sensors, and vibration sensors are deployed at corresponding spatial coordinate positions to form a distributed monitoring network with spatial correlation. Through the distributed monitoring network, each sensor is synchronously triggered to collect data according to a unified time base, thereby acquiring the mechanical deformation data, methane concentration data, and mechanical vibration data of the pipeline. The mechanical deformation data, methane concentration data, and mechanical vibration data are processed by timestamp alignment, outlier removal, and dimensional normalization to integrate and generate a multimodal sensing data set with a consistent spatiotemporal reference.

3. The underground gas pipeline safety early warning system according to claim 2, characterized in that, Humidity and temperature interference corrections were applied to the methane concentration data in the multimodal sensing dataset to obtain the corrected methane concentration data. The mechanical deformation data, mechanical vibration data, and corrected methane concentration data are fused together to obtain initial pipeline safety status assessment indicators, including: Methane concentration data is separated from the multimodal sensing data set, and ambient temperature and humidity data collected by the temperature and humidity sensor built into the methane concentration sensor during the corresponding time period are read synchronously. Based on the environmental temperature and humidity data, a preset temperature and humidity-concentration compensation relationship mapping table is queried to obtain the corresponding concentration compensation coefficient set. The concentration compensation coefficient set includes slope adjustment parameters and intercept compensation parameters corresponding to different temperature and humidity combinations. Based on the ambient temperature and humidity values ​​corresponding to the methane concentration data, the corresponding slope adjustment parameter and intercept compensation parameter are selected from the concentration compensation coefficient set; the slope adjustment parameter is multiplied by the corresponding methane concentration data to obtain a preliminary correction value; the preliminary correction value is added to the selected intercept compensation parameter to obtain the methane concentration data after environmental interference correction. Based on the mechanical deformation data and mechanical vibration data separated from the multimodal sensing data set, time domain and frequency domain feature extraction processes are performed respectively to obtain feature vectors characterizing pipeline stress state and feature vectors characterizing external disturbance intensity. The corrected methane concentration data, along with the stress state feature vector and disturbance intensity feature vector, are fused using multi-source data to obtain the initial pipeline safety status assessment index.

4. The underground gas pipeline safety early warning system according to claim 3, characterized in that, Receive initial pipeline safety status assessment indicators, delineate corresponding polygonal monitoring areas based on the topology of the distributed monitoring network, and perform grid division to obtain the area weight coefficient of each grid. Based on this weighting coefficient, a dynamic adjustment coefficient is obtained through weighted calculation. This dynamic adjustment coefficient is then used to spatially calibrate the initial pipeline safety status assessment index, resulting in a comprehensive pipeline safety status assessment index, including: Receive the initial pipeline safety status assessment index and obtain the optimal sensor node topology that constitutes the distributed monitoring network; Based on the optimal sensor node topology, the spatial coordinates of all key sensor nodes are extracted to form a set of key node spatial coordinates. For the set of key node spatial coordinates, the point with the smallest ordinate is used as the reference point, and the convex hull algorithm is performed to calculate the ordered sequence of boundary points that constitute the minimum convex boundary. The boundary points are connected sequentially according to the order of the boundary point sequence to form the minimum circumscribed convex polygon, and this polygon is determined as the corresponding polygon monitoring area. The polygonal monitoring area is divided into regular grids, and the ratio of the area of ​​each grid cell to the total area of ​​the polygon is calculated to obtain the area weight coefficient of each grid cell. Based on the area weight coefficients of each grid and combined with the distribution characteristics of historical risk data within the pre-stored grids, a dynamic adjustment coefficient is calculated using a weighted average algorithm. The initial pipeline safety status assessment index is multiplied by the dynamic adjustment coefficient to complete the spatial domain calibration process, resulting in a comprehensive pipeline safety status assessment index.

5. The underground gas pipeline safety early warning system according to claim 4, characterized in that, Based on the optimal sensor node topology, the spatial coordinates of all key sensor nodes are extracted to form a set of key node spatial coordinates. For this set, using the point with the smallest ordinate as a reference point, a convex hull algorithm is performed to calculate the ordered sequence of boundary points constituting the minimum convex boundary. The boundary points are then connected sequentially according to the order of this sequence to form a minimum circumscribed convex polygon, which is then defined as the corresponding polygon monitoring area, including: Extract the spatial coordinates of all key sensor nodes in the optimal sensor node topology to form a set of key node spatial coordinates. The point with the smallest ordinate in the set of key node spatial coordinates is identified as the reference point. If there are multiple points with the same ordinate, the point with the smallest abscissa is selected. Using the reference point as the pole, calculate the polar angle of each of the remaining points in the key node spatial coordinate set relative to the pole, and sort all the points in order of increasing polar angle to form an ordered point set; The ordered point set is traversed using a stack data structure. The turning relationship of the points is determined based on the cross product of continuous vectors. Intermediate points that do not meet the convex hull condition are removed to obtain the ordered boundary point sequence that constitutes the minimum convex boundary. Connect the boundary points sequentially according to the ordered sequence of the minimum convex boundary to form the minimum circumscribed convex polygon, and determine the polygon as the corresponding polygon monitoring area.

6. The underground gas pipeline safety early warning system according to claim 5, characterized in that, Based on the area weighting coefficients of each grid, and combined with the pre-stored historical risk data distribution characteristics within the grid, a dynamic adjustment coefficient is calculated using a weighted average algorithm, including: Receive the area weight coefficient of each grid and query the pre-stored historical risk database to obtain the historical risk level data corresponding to each grid; Based on the historical risk level data, the risk level of each grid is quantified into the corresponding risk weight coefficient through a risk level-weight mapping table. The dynamic adjustment coefficient is generated by weighting and summing the area weight coefficient of each grid with the risk weight coefficient of the corresponding grid.

7. The underground gas pipeline safety early warning system according to claim 6, characterized in that, The system receives comprehensive pipeline safety status assessment indicators, transmits them via a dual-mode communication link, and utilizes edge computing nodes to relay and buffer the transmitted data. The indicators are then successfully uploaded to the monitoring center, yielding pipeline safety status data located there, including: Receive the comprehensive pipeline safety status assessment index and encapsulate it into a data packet conforming to the dual-mode communication protocol; The data packets are sent through the primary LoRa communication link, and the link signal quality is monitored in real time to generate link status monitoring results. Based on the link status monitoring results, when the signal quality is determined to be lower than a preset threshold, a link switching command is executed to establish an effective connection with the backup NB-IoT communication link. Through the effective NB-IoT communication link, data packets are transmitted to edge computing nodes along the pipeline, where the nodes perform receiving, relaying, and buffering operations on the data packets to obtain a stable data transmission stream. The integrity of data packets that successfully arrive at the monitoring center in the transmitted data stream is verified, and valid data packets that pass the verification are selected; the valid data packets are parsed according to the protocol to extract the payload data; the payload data is reassembled according to a predefined data structure to restore and generate pipeline safety status data.

8. The underground gas pipeline safety early warning system according to claim 7, characterized in that, Dynamic risk analysis is performed on pipeline safety status data to obtain analysis results. When the analysis results exceed a preset threshold, an early warning is automatically triggered, and linkage instructions are simultaneously sent to the gas operation and maintenance platform, construction unit, and community emergency system to achieve closed-loop management, including: The pipeline safety status data is input into the risk quantification scoring model for dynamic analysis, and the multi-dimensional analysis results of the risk quantification scoring model are obtained. Based on the multi-dimensional analysis results of the risk quantification scoring model, the risk score value of the current pipeline safety status is calculated. The risk score is compared with multiple preset risk thresholds to determine the corresponding risk level; When the risk level reaches the warning threshold, a multi-level warning message and response plan matching the risk level will be automatically generated. The multi-level early warning information and response plans are simultaneously pushed to the gas operation and maintenance platform, the construction unit's operation terminal, and the community emergency system, and the response feedback information from each system is received to complete closed-loop management.

9. A control method for an underground gas pipeline network safety early warning system, characterized in that, Applied to the system as described in any one of claims 1 to 8, the method comprises: Sensor nodes are deployed at key locations on the gas pipeline and its surrounding environment to form a distributed monitoring network; the mechanical deformation, methane concentration and mechanical vibration data of the pipeline are collected synchronously through the distributed monitoring network to obtain a multimodal sensing data set; Humidity and temperature interference corrections were applied to the methane concentration data in the multimodal sensing dataset to obtain corrected methane concentration data. The mechanical deformation data, mechanical vibration data and corrected methane concentration data were then fused together to obtain the initial pipeline safety status assessment index. The system receives the initial pipeline safety status assessment index, delineates the corresponding polygonal monitoring area based on the topology of the distributed monitoring network, and performs grid division to obtain the area weight coefficient of each grid. Based on this weight coefficient, a dynamic adjustment coefficient is obtained through weighted calculation, and the initial pipeline safety status assessment index is spatially calibrated using the dynamic adjustment coefficient to obtain a comprehensive pipeline safety status assessment index. The system receives comprehensive pipeline safety status assessment indicators, transmits them through a dual-mode communication link, and uses edge computing nodes to relay and cache the transmitted data. The indicators are then successfully uploaded to the monitoring center to obtain pipeline safety status data located at the monitoring center. Dynamic risk analysis is performed on pipeline safety status data to obtain analysis results. When the analysis results exceed the preset threshold, an early warning is automatically triggered and linkage instructions are sent to the gas operation and maintenance platform, construction unit and community emergency system simultaneously to achieve closed-loop management.

10. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to perform the method as described in any one of claims 1-8.

Images