Muon imaging driven underground pipe gallery leakage detection method and system

Abstract

The invention provides an underground pipe gallery leakage detection method and system driven by muon imaging, and relates to the technical field of leakage detection, and the method comprises the steps: dividing an underground pipe gallery into N space units according to the space, and configuring a fixed array and a mobile unit to define an observation grid; executing data acquisition, and generating a muon data set through a two-way correction network; performing transmission imaging and scattering imaging inversion on the muon data set to establish a multi-scale density field, and performing time sequence residual analysis on the multi-scale density field to establish coherent abnormal data; activating a sensor group, executing data acquisition and establishing an acquisition data set; and performing data alignment on the acquired data set and the coherent abnormal data, executing a multi-source data trust fusion decision, and establishing a detection result. According to the method and the device, the technical problems of high false alarm rate and frequent missing report of leakage detection caused by insufficient monitoring capability and low hidden leakage identification precision in the prior art can be solved, and the technical effect of improving the operation and maintenance safety and the operation and maintenance efficiency of the pipe gallery is achieved.

Description

Technical Field

[0001] This application relates to the field of leakage detection technology, and in particular to a method and system for detecting leakage in underground utility tunnels driven by muon imaging. Background Technology

[0002] In the operation and maintenance of underground utility tunnels, leakage detection is a crucial step in ensuring the structural safety of the tunnel and the normal operation of its internal pipelines. Traditional leakage detection methods, such as manual inspections, require entry into the tunnel. These methods are inefficient and unsafe due to the complex environment of the tunnel, including its enclosed space, humidity, and the presence of toxic gases. Furthermore, they struggle to cover hidden areas and achieve long-term continuous monitoring. Camera-based visual inspections are susceptible to insufficient light, water mist, and structural obstructions, failing to penetrate the tunnel structure to detect internal water seepage. Conventional humidity and temperature sensors can only monitor the surface condition of the installation point, lacking the ability to detect hidden leaks caused by internal tunnel structural infiltration or soil seepage. Moreover, data from a single sensor is easily affected by environmental factors such as air pressure and temperature, leading to high false alarm or false negative rates and making it difficult to accurately identify early, minute leaks. These problems often result in the failure to detect potential leaks early, leading to serious accidents such as tunnel structural corrosion, pipeline damage, and road collapse, increasing maintenance costs and safety risks.

[0003] In summary, existing technologies suffer from insufficient monitoring capabilities and low accuracy in identifying concealed leaks, leading to high false alarm rates and frequent missed alarms in leak detection. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for detecting leakage in underground utility tunnels driven by muon imaging, in order to solve the technical problems in the prior art, such as insufficient monitoring capabilities, low accuracy in identifying concealed leakage, resulting in high false alarm rates and frequent missed alarms in leakage detection.

[0005] In view of the above problems, this application provides a method and system for detecting leakage in underground utility tunnels driven by muon imaging.

[0006] In a first aspect, this application provides a muon imaging-driven method for detecting leakage in underground utility tunnels. This method is implemented using a muon imaging-driven system for detecting leakage in underground utility tunnels. The method includes: The underground utility tunnel is divided into N spatial units. A fixed array and mobile units are configured based on these N spatial units to define an observation grid. The fixed array and mobile units are activated to perform data acquisition, and after instrument calibration via a dual-path calibration network, a muon dataset is generated. Transmission and scattering imaging inversion is performed on the muon dataset to establish a multi-scale density field. Temporal residual analysis is performed on the multi-scale density field to establish coherent anomaly data. The sensor array is simultaneously activated to perform data acquisition from the underground utility tunnel, establishing an acquisition dataset. Based on the observation grid, the acquisition dataset and the coherent anomaly data are aligned, and a multi-source data trust fusion decision is performed to establish the underground utility tunnel leakage detection results.

[0007] Secondly, this application also provides a muon imaging-driven underground utility tunnel leakage detection system for performing the muon imaging-driven underground utility tunnel leakage detection method as described in the first aspect, wherein the muon imaging-driven underground utility tunnel leakage detection system includes: The configuration module divides the underground utility tunnel into N spatial units, configures fixed arrays and mobile units based on these N spatial units, and defines an observation grid. The calibration module activates the fixed arrays and mobile units to perform data acquisition, and generates a muon dataset after instrument calibration via a dual-path calibration network. The analysis module performs transmission and scattering imaging inversion on the muon dataset to establish a multi-scale density field, performs temporal residual analysis on the multi-scale density field, and establishes coherent anomaly data. The acquisition module synchronously activates the sensor group to perform data acquisition from the underground utility tunnel and establishes an acquisition dataset. The decision module aligns the acquisition dataset and the coherent anomaly data based on the observation grid, performs multi-source data trust fusion decision-making, and establishes the underground utility tunnel leakage detection results.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: By dividing the underground utility tunnel into spatial units and configuring fixed arrays and mobile units, and combining a dual-path correction network to eliminate instrument and environmental interference to ensure the accuracy of the sub-dataset, a multi-scale density field is established through transmission-scattering imaging inversion, and coherent anomaly data is extracted through temporal residual analysis. Finally, multi-source data alignment and trust fusion decision-making are achieved based on the observation grid, realizing blind-free monitoring of the entire utility tunnel. This improves the accuracy of hidden leakage identification, reduces the false alarm rate, and shortens the early warning response time, thereby achieving the technical effect of improving the safety and efficiency of utility tunnel operation and maintenance.

[0009] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0011] Figure 1 This is a flowchart illustrating the muon imaging-driven method for detecting leakage in underground utility tunnels as described in this application. Figure 2 This is a schematic diagram of the underground pipe gallery leakage detection system driven by muon imaging according to this application.

[0012] Explanation of reference numerals in the attached diagram: Configuration module 11, Calibration module 12, Analysis module 13, Data acquisition module 14, Decision module 15. Detailed Implementation

[0013] This application provides a method and system for detecting leakage in underground utility tunnels driven by muon imaging, which solves the technical problems in the prior art, such as insufficient monitoring capabilities, low accuracy in identifying concealed leakage, resulting in high false alarm rates and frequent missed alarms.

[0014] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0015] Example 1, please refer to the appendix. Figure 1 This application provides a method for detecting leakage in underground utility tunnels driven by muon imaging. The method is applied to a muon imaging-driven underground utility tunnel leakage detection system, and specifically includes the following steps: S100: Divide the underground utility tunnel into N spatial units, configure fixed arrays and mobile units based on the N spatial units, and define an observation grid.

[0016] Specifically, firstly, based on the actual structural parameters of the underground utility tunnel, such as its total length, cross-sectional dimensions, internal compartment division, and distribution of key components, the overall three-dimensional space of the tunnel is systematically decomposed into N continuous and clearly defined spatial units. These are the basic detection subspaces determined according to the required detection accuracy and the structural characteristics of the tunnel. The value of N must balance coverage integrity and detection efficiency. For example, a 10-meter-long cuboid region matching the tunnel's cross-sectional dimensions can be designated as a single spatial unit, ensuring that each unit accurately corresponds to a local area of ​​the tunnel, providing a basic unit division for subsequent regional detection. Next, detection equipment is configured based on the divided N spatial units. For each spatial unit's tunnel sidewall, priority is given to the two sides or key internal stress-bearing sidewalls of the tunnel. A fixed array composed of muon detection modules is deployed at preset intervals. The function of this array is to collect long-term transmitted muon data within the corresponding spatial unit. By continuously monitoring the trajectory changes of muons penetrating the material within the unit, it provides data for subsequent detection. Continuous density field analysis provides stable long-term benchmark data, avoiding accidental errors in short-term detection. Simultaneously, the movement path of the maintenance robot is planned based on the distribution range of N spatial units. A mobile muon detection module is mounted on the robot to form a mobile unit. This unit, as a supplement to the fixed array, can perform dynamic scanning within the N spatial units along a preset path, focusing on covering unit corners, pipe gallery interfaces, or densely populated equipment areas that are difficult for the fixed array to reach, collecting local dynamic muon data, and filling the blind spots of fixed detection. Finally, based on the spatial boundaries and coordinate range of the N spatial units, a unified observation grid is defined. Each spatial unit is further subdivided into smaller three-dimensional grid nodes, such as a 0.5m × 0.5m × 0.5m grid. A unique spatial coordinate is assigned to each node, and the detector positions of the fixed array and the scanning trajectories of the mobile unit are mapped to this grid. This ensures that all subsequent data acquisition can be accurately associated with specific grid nodes, achieving unified spatial positioning of the data.

[0017] S200: Activate the fixed array and the mobile unit to perform data acquisition, and generate a muon dataset after performing instrument calibration through a dual-path calibration network.

[0018] Specifically, the process begins by activating the fixed array deployed on the side wall of the utility tunnel and the mobile unit carried by the maintenance robot to initiate data acquisition. In this process, the fixed array continuously captures long-term transmitted muon signals within the corresponding spatial unit based on a preset monitoring frequency to reflect the macroscopic density distribution of the utility tunnel structure. At the same time, the mobile unit scans along a pre-planned key area path, such as pipe interfaces with high leakage risk or blind spots covered by the fixed array, to dynamically acquire muon signals in local areas to supplement microscopic disturbance information. The two work together to achieve full coverage and multi-level acquisition of muon signals across the entire utility tunnel. Because the data acquisition process is susceptible to fluctuations in the instrument's own state and interference from the external environment, instrument calibration is required through a dual-path calibration network. This dual-path calibration network is a collaborative calibration system consisting of an instrument calibration channel and an environmental calibration channel, used to eliminate both instrument and environmental interference. First, the instrument calibration channel in the dual-path calibration network is activated, and a stable reference source is configured, consisting of a radioactive calibration source with a stable incident event and a reference pulse source. The acquired muon signal is compared with the reference source signal to specifically correct detector gain drift, such as small changes in detector sensitivity over time, changes in detection efficiency, and fluctuations in the detection module's response capability, as well as other instrument-related interference. The output is then optimized to eliminate interference. The first correction result for instrument interference is obtained; then the environmental correction channel is activated, and based on real-time monitored air pressure, temperature, humidity and surface disturbance parameters, the external environmental disturbance effect in the acquired signal is independently quantified and corrected, and the second correction result eliminating environmental interference is output; finally, the first correction result and the second correction result are residual matched to verify the consistency of the two types of correction results, eliminate correction deviation and priority control, such as instrument interference correction having higher priority than environmental interference correction, to ensure that core errors are eliminated first, and integrate to form a muon dataset, which is a high-quality muon signal set that has undergone dual instrument and environmental correction and can be directly used for subsequent imaging inversion.

[0019] S300: Perform transmission imaging and scattering imaging inversion on the muon dataset to establish a multi-scale density field, perform temporal residual analysis on the multi-scale density field, and establish coherent anomalous data.

[0020] Specifically, the coupled inversion framework is first activated to simultaneously perform transmission imaging and scattering imaging inversion on the muon dataset. Transmission imaging utilizes the attenuation effect caused by density differences when muons penetrate underground pipe gallery materials to generate an imaging method that reflects the macroscopic structural density distribution of the pipe gallery, which can capture large-scale density change trends. Scattering imaging inversion, on the other hand, analyzes the scattering angle shift after the interaction between muons and pipe gallery materials to invert the density perturbation in local micro-regions, which can identify small-scale structural anomalies, such as local density reduction caused by early leakage. The coupled inversion framework automatically evaluates the target scale features of the inverted target during the inversion process and constructs scale influence features. For large-scale targets such as the overall structure of the utility tunnel, it adaptively increases the weight of transmission imaging to enhance the stability of the macro density distribution. For small-scale areas prone to leakage, such as pipe interfaces and valve perimeters, it adaptively increases the weight of scattering imaging to enhance the sensitivity to local disturbances. Through this dynamic adjustment of weights, it achieves complementary fusion of transmission and scattering information. Based on the joint inversion results, a multi-scale density field is established. This density field is a density distribution model covering different spatial scales of the utility tunnel, from macroscopic, such as the overall compartment, to microscopic, such as the local pipe wall, and can fully present the density characteristics of the utility tunnel structure at different levels.

[0021] Subsequently, temporal residual analysis was performed on the multi-scale density field. First, based on the density data of each observed grid node in the multi-scale density field, a time-series density curve was constructed, recording density changes at different time points. Then, high-pass filtering was used to separate rapid density fluctuations over short periods, and low-pass filtering was used to extract slow density trends over long periods. The time-series density curve was decomposed into a short-term component reflecting sudden density changes, such as a sudden drop in local density during a leak, and a long-term component reflecting gradual density changes, such as a continuous decrease in density caused by slow leakage. Finally, the short-term and long-term components were compared with a pre-defined historical baseline model. The model is built based on density data under normal conditions of the utility tunnel, representing the density benchmark when there are no anomalies. The deviation between the two is calculated, which is the short-term disturbance residual reflecting short-term sudden anomalies and the long-term drift residual reflecting long-term gradual anomalies. Finally, based on the spatial correlation of the two types of residuals, if the same area has both short-term disturbances and long-term drift and temporal continuity, such as if the abnormal residuals continue to appear and do not disappear, they are integrated to form a coherent abnormal data set. This data set excludes accidental interference, such as instantaneous environmental fluctuations, and can reflect the temporal continuity and spatial correlation of anomalies, which can identify areas with potential leakage.

[0022] S400: Synchronously activate the sensor group to perform data acquisition in the underground utility tunnel and establish the acquisition dataset.

[0023] Specifically, based on the pre-defined observation grid, the sensor group is activated synchronously. The sensor group refers to a variety of environmental and physical signal monitoring devices that complement the muon detection. Its configuration needs to be combined with the typical characteristics of underground pipe gallery leakage, such as leakage being accompanied by local humidity increase, temperature abnormality, and water flow sound waves. Specifically, it includes humidity sensors and temperature sensors deployed at high leakage locations such as pipe interfaces, valve nodes, and low-lying water-sensitive areas on the inner wall of the pipe gallery, as well as sound wave sensors arranged at intervals along the axis of the pipe gallery to capture low-frequency signals generated by the flow or dripping of leaking water. All sensors need to be pre-coordinated with the observation grid to ensure that the collected data can be accurately mapped to specific spatial units and grid nodes.

[0024] Subsequently, data acquisition of the underground utility tunnel is carried out. This process must strictly follow the principle of time and space synchronization, that is, the data acquisition frequency and time window of the sensor group must be consistent with the data acquisition of the fixed array and the mobile unit. For example, the acquisition cycle is 5 minutes, and they are started and stopped synchronously. The acquisition content includes the real-time relative humidity value fed back by the humidity sensor, the local ambient temperature recorded by the temperature sensor, and the audio signal captured by the sound wave sensor. At the same time, the working status of each sensor is recorded in real time to avoid invalid data from being mixed in.

[0025] Finally, based on the spatial coordinates and unified timestamps of the observation grid, the multi-type data collected by the sensor group are structured and organized to establish a collection dataset. This dataset uses the grid nodes of the observation grid as the core index. Each node entry contains humidity, temperature, acoustic characteristic parameters with the corresponding timestamp, as well as sensor status identifiers, such as normal or abnormal. This forms a structured data set that combines spatial positioning, time series, and multi-dimensional parameters, ensuring that it can be accurately matched with the subsequent continuous abnormal data from the muon end.

[0026] S500: After aligning the collected dataset and the coherent abnormal data based on the observation grid, perform multi-source data trust fusion decision-making to establish underground pipe gallery leakage detection results.

[0027] Specifically, firstly, data alignment is performed based on the pre-defined observation grid. The datasets collected by the sensor group, including parameters such as humidity, temperature, and sound waves, and the coherent anomaly data obtained from the muon data through time-series residual analysis are both mapped to the three-dimensional grid nodes of the observation grid and a unified timestamp framework. Spatially, based on the grid coordinates bound during the acquisition of the two types of data, the monitoring parameters of each sensor in the acquired dataset and the residual information in the coherent anomaly data are accurately matched to the corresponding grid nodes. Temporally, based on spatiotemporal synchronous acquisition, the two types of data at the same timestamp are associated to eliminate the deviation of multi-source data in spatial location and time series, ensuring that each grid node has complete multi-source data support at any point in time.

[0028] After data alignment is completed, a multi-source data trust fusion decision is made. Based on the reliability of the data source, the anomaly score and confidence level of the collected dataset and the coherent anomaly data are determined separately. The anomaly score is calculated by a preset threshold, and the confidence level is set in combination with the characteristics of the data source. Then, the signal fusion operation formula specified in the scheme is substituted into the product operation to integrate the anomaly information of each data source, so that the data source with higher confidence level has a greater impact on the comprehensive anomaly value, thereby realizing trust-weighted multi-source information fusion.

[0029] Finally, based on the comprehensive outliers, the leakage detection results of the underground utility tunnel are established. First, the judgment threshold of the comprehensive outliers is calibrated according to the historical data of the normal operation of the utility tunnel and the leakage accident cases. The grid nodes with comprehensive outliers exceeding the threshold are marked as suspected leakage areas. Combining the short-term / long-term residual characteristics of the continuous outlier data and the parameter anomaly type of the collected dataset, the severity of leakage, leakage type and corresponding actual physical location of the utility tunnel in the suspected area are further marked. Finally, a structured detection result containing spatial location, temporal characteristics and severity is formed. The result is a visualized leakage distribution map and a detailed text report.

[0030] Furthermore, after performing instrument calibration via a dual-path calibration network, the generated muon dataset includes: Activate the instrument calibration channel in the dual-path calibration network, configure a stable reference source, which provides a radioactive calibration source and reference pulse source for stable incident events, and perform detector gain drift and detection efficiency change correction in the data acquisition results based on the stable reference source, outputting a first calibration result to eliminate instrument interference; activate the environmental calibration channel in the dual-path calibration network, independently correct for external disturbance effects in the data acquisition results based on air pressure, temperature, humidity, and surface disturbance parameters, and output a second calibration result to eliminate environmental factor interference; perform residual matching and priority control on the first and second calibration results to establish a muon dataset.

[0031] Specifically, the instrument calibration channel in the dual-path calibration network, responsible for eliminating the detector's own errors, is activated first. This channel is one of the functional modules of the dual-path calibration network, precisely calibrating the muon detector to address performance fluctuations that occur during long-term operation. Simultaneously, a stable reference source is configured to provide a standard reference signal, serving as the basis for instrument calibration. This stable reference source contains two key components: a radioactive calibration source to provide stable incident events (muon incident signals with constant intensity, direction, and energy), which can be used as a standard template in the calibration process to ensure consistency of calibration results; and a reference pulse source to continuously release radioactive particle signals of fixed intensity, whose signal characteristics are highly similar to muon signals, simulating the stable incident state of real muons. Finally, a reference pulse source is used to calibrate the detector's signal response. This device outputs electrical pulse signals with fixed amplitude and frequency, directly verifying the detector's ability to receive and convert standard signals. Based on the standard signal provided by the stable reference source, the raw muon data collected by the fixed array and the moving unit are compared point by point to specifically correct two types of core instrument errors: First, detector gain drift, which refers to the slight shift in the amplitude of the output electrical signal as the detector receives the same intensity of incident signal over time, causing data intensity distortion. The signal amplification coefficient of the detector is calibrated by the fixed amplitude signal of the reference pulse source to restore gain stability. Second, detection efficiency variation, which refers to the fluctuation of the proportion of incident muon signals successfully captured and recorded by the detector with environmental factors or usage time, causing data quantity deviation. The signal capture rate of the detector is statistically analyzed by the stable incident events of the radioactive calibration source to correct the impact of efficiency decay. Finally, the first correction result is output, eliminating the instrument's own interference and retaining only the true muon signal and environmental interference.

[0032] After instrument calibration is completed, the environmental calibration channel in the dual-path calibration network, which is responsible for removing the influence of the external environment, is activated. The environmental calibration channel is another module of the dual-path calibration network, used to eliminate the interference of changes in the external environment of the utility tunnel on the muon data. Based on the real-time monitored disturbance parameters, that is, the set of external environmental factors that affect the accuracy of muon signal acquisition, specifically including air pressure, temperature, humidity and surface disturbance. Among them, changes in air pressure will change the propagation path of muons in the air, temperature and humidity will affect the electrical properties of detector materials, and surface disturbances such as construction vibrations around the utility tunnel may cause slight displacement of the detector position. Then, the external disturbance effects in the original data are independently calibrated. The external disturbance effects are false signals or signal deviations caused by disturbance parameters and superimposed on the real muon signal. For example, increased humidity may increase detector noise. Independent calibration means that without relying on the instrument calibration results, a quantitative model of environmental factors and signal deviation is established separately. Environmental interference is removed by reverse calculation to avoid mutual influence of different interference sources. Finally, the second calibration result is output, which eliminates the interference of environmental factors and retains only the real muon signal and the instrument error has been eliminated.

[0033] Finally, residual matching is performed on the first and second correction results. Residual matching refers to calculating the data deviation value of the two types of correction results under the same spatiotemporal coordinates to verify the consistency of the correction logic. If the deviation exceeds the preset threshold, recalibration is performed to ensure that there is no contradiction between the two types of correction results. Priority control is also performed, which sets the weight of the correction results according to the degree of impact of the error on the data quality. Since the impact of instrument error on data accuracy is more direct and less controllable, the first correction result has a higher priority than the second correction result. When there is a small deviation between the two types of results, the environmental compensation information of the second correction result is fused based on the first correction result. Through data integration, a muon dataset is formed. This muon dataset is a high-quality dataset that has undergone dual interference removal by instruments and environment and signal feature standardization. Its data can truly reflect the propagation and interaction characteristics of muons in underground pipe corridors and can be directly used for subsequent transmission imaging and scattering imaging inversion.

[0034] Furthermore, the step of performing transmission and scattering imaging inversion on the muon dataset to establish a multi-scale density field includes: The coupled inversion framework is activated to perform joint inversion of the muon dataset using transmission imaging and scattering imaging. The coupled inversion framework is used to adaptively adjust the transmission imaging weight and scattering imaging weight. During the inversion process, the target scale feature evaluation of the inverted target is performed to establish scale influence features. Based on the scale influence features, large-scale stable transmission enhancement and small-scale local perturbation enhancement are matched to adaptively adjust the transmission imaging weight and scattering imaging weight. A multi-scale density field is established based on the joint inversion results.

[0035] Specifically, the first step is to activate the coupled inversion framework, which is an algorithmic framework that integrates target scale analysis, adaptive control of transmission-scattering weights, and fusion of multiple imaging results. It includes a target scale feature evaluation module, a weight calculation module, and a result integration module. The core logic is to break through the resolution limitations of a single imaging technology through modular collaboration and achieve accurate extraction of macroscopic and microscopic density information. After activating the framework, joint inversion of transmission imaging and scattering imaging is performed on the muon dataset. Transmission imaging is a technique for inverting the density distribution of large-scale structures in pipe corridors, such as entire chambers and large pipe bodies, based on the signal attenuation effect caused by density differences when muons penetrate the material of the pipe corridor. Following the Lambert-Beer law, the attenuation coefficient is positively correlated with the material density. It is good at capturing large-scale, stable density changes. Scattering imaging, on the other hand, analyzes the scattering angle shift after the interaction between muons and the atomic nuclei of the material. Based on quantum scattering theory, the shift is positively correlated with the amplitude of local density perturbation. It is a technique for inverting density anomalies in small-scale regions, such as pipe joints and valve seals, and is more sensitive to minute perturbations at the millimeter to centimeter level. Joint inversion is driven by the framework algorithm to process the raw signals of the two imaging methods in parallel. Through real-time data interaction between modules, the macroscopic density trend and microscopic perturbation information are complemented.

[0036] During the joint inversion process, the framework first clarifies the inversion target, namely the structural units within the utility tunnel that need to be analyzed, covering different levels of objects such as large-scale overall compartments, medium-scale single pipes, and small-scale pipe interfaces / valves. Then, the target scale feature evaluation module performs target scale feature evaluation, extracting three parameters of the inversion target algorithmically: spatial size threshold (comparing the actual target size with the preset scale division standard); structural complexity (analyzing the utility tunnel CAD model to determine if there are complex structures prone to leakage, such as splicing gaps); and historical density change rate (retrieving the density baseline data of the utility tunnel under normal conditions to determine the probability of density fluctuations in the target area), and calculating the target scale feature value according to a weighted formula of spatial size, structural complexity, and historical change rate. The specific formula is as follows: ; The target scale feature value quantifies the scale attributes and leakage sensitivity potential of the target. The value ranges from 0 to 10. The lower the value, the more the target is biased towards large scale and low leakage sensitivity, and the higher the value, the more it is biased towards small scale and high leakage sensitivity. The spatial size is scored from 0 to 10, based on the quantification of the actual spatial size of the inverted target. The smaller the size, the lower the score, and the larger the size, the higher the score. Spatial size weights are assigned based on pre-defined scheme parameters; The structural complexity score is based on whether the target contains easily leaking structures such as splicing gaps and seals, ranging from 0 to 10 points. The more complex the structure and the more easily leaking components there are, the higher the score. The structural complexity weight is based on the pre-set scheme; H is the historical density change rate score, which is quantified based on whether there are density anomaly records in the target's history, ranging from 0 to 10 points. The more frequent the historical anomalies and the greater the change, the higher the score. The weighting is based on the historical density change rate, pre-set in the scheme; and ensures... + + =1.

[0037] After completing the feature evaluation, scale influence features are established based on the target scale feature values. These features are mathematical models that quantify the sensitivity of the inverted target to transmission and scattering imaging. First, sensitivity coefficients for the two types of imaging are determined, ranging from 0 to 1. Higher coefficients indicate a more significant response of the imaging technology to changes in target density. Specifically, if the target scale feature value is ≤ 5, a transmission sensitivity coefficient of 0.8 and a scattering sensitivity coefficient of 0.2 are set, reflecting that this type of target is more suitable for transmission imaging and adept at capturing macroscopic density changes. If the feature value is > 5, a transmission sensitivity coefficient of 0.3 and a scattering sensitivity coefficient of 0.7 are set, reflecting that it is more suitable for scattering imaging and adept at capturing microscopic density disturbances. The sensitivity coefficients can be fine-tuned according to the feature values.

[0038] Based on the scale-related characteristics, the framework performs targeted enhancement adjustments through a weight calculation module. For large-scale inversion targets, it performs large-scale stable transmission enhancement. The algorithm dynamically increases the transmission imaging weight from the default value of 0.5 to 0.8, which matches the sensitivity coefficient, while reducing the scattering imaging weight to 0.2. By increasing the transmission weight, the stability of the macroscopic density distribution is enhanced, avoiding interference from small-scale scattering signals on macroscopic data. For small-scale inversion targets, it performs small-scale local perturbation enhancement matching. The algorithm increases the scattering imaging weight to 0.7 and reduces the transmission imaging weight to 0.3, based on the scattering sensitivity coefficient. At the same time, a local signal amplification algorithm is introduced to perform a 2x gain processing on signals whose scattering angle offset exceeds the normal threshold, ensuring that small-scale density perturbations are not masked by macroscopic density signals.

[0039] After weight adjustment and signal processing, the framework, through the result integration module, precisely matches the macroscopic density data obtained from transmission imaging and the microscopic density data obtained from scattering imaging according to the pre-defined three-dimensional coordinates of the observation grid and the data acquisition timestamps, generating a joint inversion result. This result is a structured dataset, with each grid node as a unit, containing the density value corresponding to the node, its scale level, and the reliability of the density data. Finally, based on the joint inversion result and combined with the spatial topological relationship of the observation grid, a multi-scale density field is constructed. The multi-scale density field is a three-dimensional structured density model, divided into different accuracy levels according to scale levels, which can intuitively present the density distribution characteristics of the utility tunnel from the whole to the local, providing a hierarchical density benchmark for subsequent anomaly analysis.

[0040] Furthermore, the step of performing time-series residual analysis on the multi-scale density field to establish coherent anomaly data includes: A time series density curve is established based on the multi-scale density field; the time series density curve is decomposed by high-pass filtering and low-pass filtering to output short-term components and long-term components; the short-term components and the long-term components are compared with the historical baseline model respectively, and the short-term disturbance residual and the long-term drift residual are calculated respectively; coherent anomalous data is established based on the short-term disturbance residual and the long-term drift residual.

[0041] Specifically, each three-dimensional grid node in the multi-scale density field observation grid is first used as an independent analysis unit. Based on density monitoring data at fixed time intervals, such as every 5 minutes, a time series density curve is constructed in chronological order. This curve is the density change trajectory of each grid node in a continuous time period, which can intuitively reflect the dynamic fluctuation of density at the corresponding location over time. For example, the density curve of a pipe interface node in a utility tunnel can record its density change trend from normal state to the initial stage of leakage, providing a time dimension data basis for subsequent anomaly analysis.

[0042] Next, high-pass and low-pass filtering are performed simultaneously on each time-series density curve to complete the frequency decomposition of the signal. The high-pass filter sets a high-frequency signal retention threshold, retaining only density fluctuations with a period less than one hour, filtering out slowly changing long-term signals in the curve, and finally outputting the short-term component. This component mainly corresponds to sudden density changes within a short period, such as a rapid drop in density caused by rapid infiltration of local water due to a pipe rupture, such as a drop from 2.5 g / cm³ within 10 minutes. 3 Reduced to 2.3 g / cm³ 3 The short-term fluctuations caused by instantaneous electromagnetic interference from the detector can be filtered out by setting a threshold for retaining low-frequency signals, such as retaining only density changes with a period greater than 24 hours. The long-term component is output, which mainly corresponds to the gradual density changes over a long period of time, such as the continuous decrease in density caused by minor leakage due to the aging of the sealing gasket, or the slow density drift caused by the natural settlement of the pipe gallery structure. This decomposition can effectively distinguish density changes caused by different reasons and avoid confusing the signals of sudden leakage with those of slow leakage.

[0043] The decomposed short-term and long-term components are then compared point-by-point with a pre-defined historical baseline model. This historical baseline model is based on multi-scale density field data from the normal operation phase of the utility tunnel, without leakage or structural anomalies. It establishes a density benchmark model for each grid node through statistical analysis, such as mean and standard deviation calculations. This model includes short-term fluctuation benchmarks under normal conditions, such as the range of short-term density deviations caused by normal environmental fluctuations, and long-term trend benchmarks, such as the range of long-term density drift caused by normal temperature and humidity changes. These are the core references for determining the presence of anomalies. During the comparison, the short-term disturbance residual is calculated, which is the absolute value of the difference between the current short-term component and the corresponding short-term benchmark in the historical baseline model. If this value exceeds the normal fluctuation threshold, it indicates a sudden short-term density anomaly, possibly corresponding to sudden leakage. The long-term drift residual is the absolute value of the difference between the current long-term component and the corresponding long-term benchmark in the historical baseline model. If this value exceeds the normal drift threshold, it indicates a gradual long-term density anomaly, possibly corresponding to slow leakage. The time points and grid node locations where both types of residuals exceed the thresholds are recorded.

[0044] Finally, based on the spatiotemporal correlation between short-term disturbance residuals and long-term drift residuals, coherent anomalous data is integrated to form a coherent set of data. Specifically, grid nodes with short-term disturbance residuals exceeding the threshold and accompanied by long-term drift residuals exceeding the threshold, or with short-term residuals exceeding the threshold multiple times consecutively, are selected. Random interference such as transient electromagnetic interference, which only causes a single instance of short-term residual exceeding the threshold, is excluded. At the same time, the anomalous residual values, anomalous durations, and spatial coordinates of these nodes are structurally integrated with their corresponding observation grid locations to form an anomalous dataset that combines temporal continuity and spatial correlation, i.e., continuous occurrence of anomalous events and concentrated anomalous events in specific areas. This ensures that the data can accurately point to the density anomalies caused by actual leakage, rather than irrelevant interference.

[0045] Furthermore, the process of performing multi-source data trust fusion decision-making and establishing underground utility tunnel leakage detection results includes: Based on the anomaly scores and confidence levels of the multi-source data, a signal fusion operation is performed as follows: ; in, Characterized in position ,time The overall outlier, Represents the total number of data from multiple sources. Indexes that represent data from any multiple sources For the first Confidence level of the source data Characterizing the first Source data in location ,time The abnormal scores are calculated; the leakage detection results of the underground utility tunnel are established based on the signal fusion calculation results.

[0046] Specifically, the first step is to clarify the total number of multi-source data participating in the fusion. This parameter represents the total number of independent data sources participating in the signal fusion operation. It also assigns an index to each data source based on the number of data sources involved. Used to uniquely distinguish different data sources, for example, to mark consecutive outlier data as... =1, the collected dataset is labeled as =2, the index can accurately locate the abnormal scores and confidence levels of each data source, avoiding data confusion.

[0047] Then for each index Anomaly scores were calculated for each corresponding data source. With the Confidence of source data Among them, the abnormal score It is the first quantification Source data in a specific location ,time The parameter for the degree of abnormality ranges from 0 to 1, where 0 indicates no abnormality and 1 indicates extreme abnormality. For continuous outlier data with a value of 1, the score is calculated based on the deviation magnitude of its short-term disturbance residuals, long-term drift residuals, and historical baseline model. The formula is (actual residual - normal deviation threshold) / normal deviation threshold. If the residual does not exceed the threshold, the score is 0. For the dataset with a value of 2, the following calculations are performed for parameters such as humidity, temperature, and sound waves: (actual parameter value - upper limit of normal range) / upper limit of normal range. This calculation is only performed when a parameter exceeds its limit; otherwise, it is 0. The maximum value of each parameter is then taken as the anomaly score for that data source, ensuring the capture of the most significant environmental anomalies. The confidence score for the i-th source data is then determined. It reflects the first The weighting parameter for the reliability of the source data, with a value range of 0-1, where 1 indicates complete reliability.

[0048] After completing the parameter calculation, substitute the parameters into the signal fusion calculation formula specified in the scheme to perform the calculation: the formula first passes through " "Calculate the non-anomaly probability of a single data source, that is, the probability that the data source is determined to have no leakage, and then perform a product operation." Integrate non-anomaly information from all data sources, and finally subtract the product from 1 to obtain the location. ,time Comprehensive outliers under This value is a quantitative result of the overall anomaly level after integrating multi-source data. It ranges from 0 to 1; a higher value indicates a greater probability of leakage at the corresponding spatiotemporal point. For example... When the value is 0.8, it indicates that the possibility of leakage at this location is extremely high.

[0049] Finally, based on the comprehensive outlier values To establish a system for detecting leaks in underground utility tunnels, the threshold for identifying comprehensive anomalies is first calibrated based on normal operational data and historical leakage cases. Then, the observation grid is used to... Grid nodes with a value ≥0.6 are marked as suspected leakage areas, and details are supplemented based on the anomaly characteristics of each data source, such as... =1 indicates a long-term drift residual anomaly. =2 When there is a continuous increase in humidity, it is marked as a long-term slow leakage; =1 indicates a short-term disturbance that causes a sharp increase in residuals. =2 When there is an abnormal sound wave, it is marked as a short-term sudden leakage. At the same time, the grid coordinates are converted into the actual mileage and height of the pipe gallery, the physical location of the suspected leakage area is marked, and the severity is divided according to the comprehensive anomaly value: 0.6-0.7 is mild, 0.7-0.8 is moderate, and ≥0.8 is severe. Finally, the detection results are generated, which include a visualized heat map of leakage distribution and a structured text report.

[0050] Furthermore, the fixed array is deployed on the side wall of the utility tunnel to collect long-term transmitted muon data, and the mobile unit is a mobile muon detection unit, which is carried by the maintenance robot to perform dynamic scanning within a predetermined area.

[0051] Specifically, the fixed array is a static monitoring device consisting of multiple muon detection modules. Deployed on the sidewalls of the utility tunnel, and using the three-dimensional coordinates of the observation grid, the detection modules are fixed at 1-2 meter intervals on both sides of each spatial unit, avoiding obstructions such as pipe supports and cables. This ensures that the module's detection range completely covers the cross-section and longitudinal extension area of ​​the corresponding spatial unit, and that all modules' signal acquisition ports face inwards to capture muon signals penetrating the tunnel structure. Once deployed, the fixed array's function is to collect long-term transmitted muon data. This data refers to muon penetration signals recorded continuously for 24 hours with a 5-10 minute acquisition cycle. It reflects changes in the macroscopic density distribution of the utility tunnel structure over a longer period, such as the slow decrease in density in localized areas due to leakage. Because the acquisition cycle is stable and the coverage is fixed, it serves as the basic reference data for subsequent judgment of structural anomalies, avoiding interference from short-term fluctuations.

[0052] Simultaneously, a mobile unit is configured. The core of this unit is a mobile muon detection unit, smaller than the fixed detection module and a lightweight detection device. It features wireless signal transmission capabilities and is compatible with robot mounting requirements. It is fixedly installed on the top detection bracket of a dedicated pipe gallery maintenance robot. The maintenance robot has autonomous navigation, track-based movement, or wheeled movement capabilities, allowing it to move autonomously within the pipe gallery according to a preset path. It also carries a data storage and real-time communication module, enabling synchronous transmission of detection data. This ensures that the signal reception direction of the detection unit is consistent with the fixed array, forming a complementary data acquisition angle. Subsequently, based on the leakage risk level of N spatial units, such as units with pipe interfaces, units in the fixed array's blind spot, and historical leakage units—areas with higher risk—the maintenance robot... The human control system pre-determines a set of spatial units that need to be monitored, typically covering 30%-50% of the total length of the utility tunnel. Each pre-determined area is marked with the corresponding observation grid coordinate range. After the robot, equipped with a mobile muon detection unit, enters the pre-determined area, it starts dynamic scanning and moves along the axis of the utility tunnel at a speed of 2 meters per minute. At the same time, it shortens the acquisition cycle to 1-2 minutes and performs high-density, close-range muon signal acquisition on the utility tunnel structure within the pre-determined area. This can not only supplement the blind spots of the fixed array in local details, but also capture short-term, local density disturbance signals through dynamic scanning, such as a small-scale density drop caused by a sudden leak. This forms a macro-micro, static-dynamic acquisition complement to the long-term data of the fixed array.

[0053] Furthermore, after establishing the leakage detection results of the underground utility tunnel, the following is included: Based on the leakage detection results of the underground utility tunnel, leakage early warning matching is performed to establish a leakage early warning signal; after an abnormality is reported in the leakage early warning signal, continuous monitoring of the leakage location is performed, and the early warning response cycle is dynamically adjusted to perform response monitoring of the leakage early warning signal.

[0054] Specifically, the process begins by matching leak warnings based on the detection results. This involves comparing key parameters from the detection results, such as comprehensive anomalies, leak duration, and density change rate, with a pre-defined warning level standard library. This library is pre-set according to the utility tunnel operation and maintenance specifications. For example, a comprehensive anomaly of 0.6-0.7 corresponds to a yellow warning, 0.7-0.8 to an orange warning, and ≥0.8 to a red warning. Since short-term, sudden leaks may spread rapidly, the warning level is raised by one level based on the matching. The warning level for each leak is determined through precise parameter matching, thereby establishing a leak warning signal. This signal is a structured information package containing the warning level, the actual physical location of the leak, the leak type, the first detection time, and the real-time comprehensive anomaly value. It can be directly connected to the utility tunnel operation and maintenance management platform to achieve standardized storage and transmission of warning information.

[0055] After the leakage early warning signal is constructed, anomaly reporting is implemented. This involves simultaneously pushing the early warning signal to maintenance personnel's terminals via the multi-channel output module of the utility tunnel operation and maintenance system, such as mobile app pop-ups and SMS alerts. The content includes the warning level, leakage location, and information about the central control room's audible and visual alarm devices. Different warning levels trigger different frequencies of audible and visual prompts; a red warning triggers high-frequency flashing lights and a rapid alarm sound. The maintenance management platform also marks the leakage location on the utility tunnel's electronic map with different colored icons. Clicking on an icon displays the complete warning information, ensuring the maintenance team receives and responds to the warning immediately. Simultaneously, continuous monitoring is initiated at the leakage location. This involves increasing the frequency of the fixed array's data collection in the corresponding area, for example, from the usual 5 minutes / time to 1 minute / time. A maintenance robot equipped with a mobile unit is deployed to perform fixed-point cyclic scanning of the leakage location, covering the leakage area at fixed intervals. This focuses on capturing local density changes and humidity fluctuations, updating leakage parameters in real time, and providing dynamic data support for adjusting the early warning response.

[0056] During continuous monitoring, the early warning response cycle is dynamically adjusted based on the real-time collected leakage parameter change trends. This cycle refers to the interval between the maintenance system's review of the leakage status and the on-site response time of maintenance personnel. For example, the initial response cycle for a yellow warning is set to 2 hours. If continuous monitoring detects that the leakage is worsening, the response cycle is shortened to 30 minutes, and an emergency on-site response command from maintenance personnel is triggered. If the leakage is detected to be stable or alleviated, the response cycle is extended to 4 hours to avoid over-response and resource waste. Finally, response monitoring is performed based on the dynamically adjusted early warning response cycle, that is, the progress of the handling of the early warning signal is continuously tracked, including whether maintenance personnel arrive on-site on time, whether the leakage is effectively controlled, and whether the monitoring parameters match the early warning level. If the response monitoring finds that the leakage has not been alleviated or the deviation between the early warning signal and the actual situation exceeds the threshold, a second early warning review is triggered to ensure that the leakage early warning forms a complete process from signal generation - anomaly reporting - continuous monitoring - cycle adjustment - effect tracking.

[0057] In summary, the muon imaging-driven underground utility tunnel leakage detection method provided in this application has the following technical advantages: By dividing the underground utility tunnel into spatial units and configuring fixed arrays and mobile units, and combining a dual-path correction network to eliminate instrument and environmental interference to ensure the accuracy of the sub-dataset, a multi-scale density field is established through transmission-scattering imaging inversion, and coherent anomaly data is extracted through temporal residual analysis. Finally, multi-source data alignment and trust fusion decision-making are achieved based on the observation grid, realizing blind-free monitoring of the entire utility tunnel. This improves the accuracy of hidden leakage identification, reduces the false alarm rate, and shortens the early warning response time, thereby achieving the technical effect of improving the safety and efficiency of utility tunnel operation and maintenance.

[0058] Example 2: Based on the same inventive concept as the muon imaging-driven underground utility tunnel leakage detection method in Example 1, this application also provides a muon imaging-driven underground utility tunnel leakage detection system. Please refer to the appendix. Figure 2 The system includes: Configuration module 11 divides the underground utility tunnel into N spatial units, configures a fixed array and a moving unit based on the N spatial units, and defines an observation grid; Correction module 12 activates the fixed array and the moving unit to perform data acquisition, and generates a muon dataset after performing instrument correction through a dual-path correction network; Analysis module 13 performs transmission imaging and scattering imaging inversion on the muon dataset to establish a multi-scale density field, performs time-series residual analysis on the multi-scale density field, and establishes coherent anomaly data; Acquisition module 14 synchronously activates the sensor group to perform data acquisition of the underground utility tunnel and establishes an acquisition dataset; Decision module 15 performs data alignment on the acquisition dataset and the coherent anomaly data based on the observation grid, performs multi-source data trust fusion decision, and establishes the underground utility tunnel leakage detection results.

[0059] Furthermore, the calibration module 12 in the muon imaging-driven underground utility tunnel leakage detection system is used for: Activate the instrument calibration channel in the dual-path calibration network, configure a stable reference source, which provides a radioactive calibration source and reference pulse source for stable incident events, and perform detector gain drift and detection efficiency change correction in the data acquisition results based on the stable reference source, outputting a first calibration result to eliminate instrument interference; activate the environmental calibration channel in the dual-path calibration network, independently correct for external disturbance effects in the data acquisition results based on air pressure, temperature, humidity, and surface disturbance parameters, and output a second calibration result to eliminate environmental factor interference; perform residual matching and priority control on the first and second calibration results to establish a muon dataset.

[0060] Furthermore, the analysis module 13 in the muon imaging-driven underground utility tunnel leakage detection system is used for: The coupled inversion framework is activated to perform joint inversion of the muon dataset using transmission imaging and scattering imaging. The coupled inversion framework is used to adaptively adjust the transmission imaging weight and scattering imaging weight. During the inversion process, the target scale feature evaluation of the inverted target is performed to establish scale influence features. Based on the scale influence features, large-scale stable transmission enhancement and small-scale local perturbation enhancement are matched to adaptively adjust the transmission imaging weight and scattering imaging weight. A multi-scale density field is established based on the joint inversion results.

[0061] The analysis module 13 is also used for: A time series density curve is established based on the multi-scale density field; the time series density curve is decomposed by high-pass filtering and low-pass filtering to output short-term components and long-term components; the short-term components and the long-term components are compared with the historical baseline model respectively, and the short-term disturbance residual and the long-term drift residual are calculated respectively; coherent anomalous data is established based on the short-term disturbance residual and the long-term drift residual.

[0062] Furthermore, the decision module 15 in the muon imaging-driven underground utility tunnel leakage detection system is used for: Based on the anomaly scores and confidence levels of the multi-source data, a signal fusion operation is performed as follows: ; in, Characterized in position ,time The overall outlier, Represents the total number of data from multiple sources. Indexes that represent data from any multiple sources For the first Confidence level of the source data Characterizing the first Source data in location ,time The abnormal scores are calculated; the leakage detection results of the underground utility tunnel are established based on the signal fusion calculation results.

[0063] Furthermore, the configuration module 11 in the muon imaging-driven underground utility tunnel leakage detection system is used for: A fixed array is deployed on the side wall of the utility tunnel to collect long-term transmitted muon data. The mobile unit is a mobile muon detection unit, which is carried by a maintenance robot to perform dynamic scanning within a predetermined area.

[0064] Furthermore, the decision module 15 in the muon imaging-driven underground utility tunnel leakage detection system is used for: After establishing the leakage detection results of the underground utility tunnel, the process includes: matching leakage early warnings based on the leakage detection results of the underground utility tunnel, and establishing a leakage early warning signal; after an abnormal report of the leakage early warning signal is issued, continuous monitoring of the leakage location is performed, and the early warning response cycle is dynamically adjusted to perform response monitoring of the leakage early warning signal.

[0065] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1The muon imaging-driven underground utility tunnel leakage detection method and specific examples in Embodiment 1 are also applicable to the muon imaging-driven underground utility tunnel leakage detection system of this embodiment. Through the foregoing detailed description of the muon imaging-driven underground utility tunnel leakage detection method, those skilled in the art can clearly understand the muon imaging-driven underground utility tunnel leakage detection system of this embodiment. Therefore, for the sake of brevity, it will not be described in detail here. As the system disclosed in the embodiments corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant details can be found in the method section.

[0066] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0067] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for detecting leakage in underground utility tunnels driven by muon imaging, characterized in that, The method includes: The underground utility tunnel is divided into N spatial units, and a fixed array and a mobile unit are configured based on the N spatial units to define an observation grid. The fixed array and the mobile unit are activated to perform data acquisition, and after instrument calibration is performed through a dual-path calibration network, a sub-dataset is generated. Transmission and scattering imaging inversions are performed on the muon dataset to establish a multi-scale density field. Temporal residual analysis is then performed on the multi-scale density field to establish coherent anomalous data. Synchronously activate the sensor group to collect data from the underground utility tunnel and establish a dataset. After aligning the collected dataset and the coherent abnormal data based on the observation grid, a multi-source data trust fusion decision is performed to establish the underground utility tunnel leakage detection results.

2. The muon imaging-driven method for detecting leakage in underground utility tunnels as described in claim 1, characterized in that, After performing instrument calibration via a dual-path calibration network, a muon dataset is generated, including: Activate the instrument calibration channel in the dual-path calibration network, configure a stable reference source, which is used to provide a radioactive calibration source and a reference pulse source for stable incident events, and perform detector gain drift and detection efficiency change correction in the data acquisition results based on the stable reference source, and output the first calibration result to eliminate instrument interference; Activate the environmental correction channel in the dual-path correction network, independently correct the external disturbance effect in the data acquisition results based on air pressure, temperature, humidity and surface disturbance parameters, and output the second correction result that eliminates the interference of environmental factors; The first and second correction results are subjected to residual matching and priority control to establish the Mu sub-dataset.

3. The muon imaging-driven method for detecting leakage in underground utility tunnels as described in claim 1, characterized in that, The process of performing transmission and scattering imaging inversion on the muon dataset to establish a multi-scale density field includes: The coupled inversion framework is activated to perform joint inversion of transmission imaging and scattering imaging on the muon dataset. The coupled inversion framework is used to adaptively adjust the transmission imaging weight and scattering imaging weight. During the inversion process, the target scale feature evaluation of the inverted target is performed to establish scale influence features. Based on the scale influence features, large-scale stable transmission enhancement and small-scale local perturbation enhancement matching are performed to adaptively adjust the transmission imaging weight and scattering imaging weight. A multi-scale density field is established based on the joint inversion results.

4. The muon imaging-driven method for detecting leakage in underground utility tunnels as described in claim 3, characterized in that, The step of performing time-series residual analysis on the multi-scale density field to establish coherent anomaly data includes: A time-series density curve is established based on the multi-scale density field; The time series density curve is decomposed by high-pass filtering and low-pass filtering to output short-term and long-term components. The short-term component and the long-term component are compared with the historical baseline model respectively, and the short-term disturbance residual and the long-term drift residual are calculated respectively. Based on the short-term disturbance residuals and long-term drift residuals, coherent anomaly data is established.

5. The muon imaging-driven method for detecting leakage in underground utility tunnels as described in claim 1, characterized in that, The process of performing multi-source data trust fusion decision-making and establishing underground utility tunnel leakage detection results includes: Based on the anomaly scores and confidence levels of the multi-source data, a signal fusion operation is performed as follows: ； in, Characterized in position ,time The overall outlier, Represents the total number of data from multiple sources. Indexes that represent data from any multiple sources For the first Confidence level of the source data Characterizing the first Source data in location ,time Abnormal scores below; The leakage detection results of underground utility tunnels are established based on the signal fusion calculation results.

6. The muon imaging-driven method for detecting leakage in underground utility tunnels as described in claim 1, characterized in that, The fixed array is deployed on the side wall of the utility tunnel to collect long-term transmitted muon data, and the mobile unit is a mobile muon detection unit, which is carried by a maintenance robot to perform dynamic scanning within a predetermined area.

7. The muon imaging-driven method for detecting leakage in underground utility tunnels as described in claim 1, characterized in that, After establishing the leakage detection results of the underground utility tunnel, the following are included: Based on the leakage detection results of the underground utility tunnel, leakage early warning matching is performed to establish a leakage early warning signal; After an abnormal report of a leakage warning signal is issued, the leakage location is continuously monitored, and the warning response cycle is dynamically adjusted to monitor the response to the leakage warning signal.

8. A muon-based imaging-driven underground utility tunnel leakage detection system, characterized in that, The system is used for implementing the muon imaging-driven underground utility tunnel leakage detection method according to any one of claims 1 to 7, wherein the system comprises: The configuration module divides the underground utility tunnel into N spatial units, configures fixed arrays and mobile units based on the N spatial units, and defines an observation grid. The calibration module activates the fixed array and the mobile unit to perform data acquisition, and after performing instrument calibration through a dual-path calibration network, generates a sub-data set. The analysis module performs transmission and scattering imaging inversion on the muon dataset, establishes a multi-scale density field, performs temporal residual analysis on the multi-scale density field, and establishes coherent anomaly data. The data acquisition module synchronously activates the sensor group to perform data acquisition from the underground utility tunnel and establishes the acquisition dataset. The decision module aligns the collected dataset and the coherent abnormal data based on the observation grid, performs multi-source data trust fusion decision-making, and establishes the underground pipe gallery leakage detection results.

Images