Underground pipe gallery intelligent inspection robot and hidden danger dynamic identification system

Abstract

The present application relates to the technical field of underground pipe gallery intelligent inspection, and discloses an underground pipe gallery intelligent inspection robot and a hidden danger dynamic identification system.The system comprises an environment perception calibration module, a hidden danger feature extraction module and an identification strategy optimization module.The environment perception calibration module verifies the data perception accuracy to obtain pipe gallery perception indexes, and judges whether to execute hidden danger feature extraction; the hidden danger feature extraction module evaluates the extraction effect to obtain pipe gallery feature indexes, and judges whether to execute strategy optimization; and the identification strategy optimization module obtains pipe gallery optimization indexes, and judges whether to execute inspection resource dynamic priority configuration.The system comprises environment data collection verification, hidden danger feature extraction evaluation and identification strategy optimization judgment units, and the functions correspond to the robot.The present application can improve data collection accuracy, optimize hidden danger feature extraction and identification strategy, realize inspection resource dynamic configuration, and guarantee the safe operation of underground pipe galleries.

Description

Technical Field

[0001] This invention relates to the field of intelligent inspection technology for underground utility tunnels, specifically to an intelligent inspection robot for underground utility tunnels and a dynamic hazard identification system. Background Technology

[0002] With the continuous development of urban infrastructure construction, underground utility tunnels, as an important urban infrastructure, undertake the centralized laying of various pipelines such as electricity, communication, gas, and water supply and drainage. Their operational safety is directly related to the normal operation of the city and the safety of residents' lives and property. However, the underground utility tunnel environment is complex, with problems such as high humidity, enclosed space, and potential for harmful gases. Traditional manual inspection methods are not only inefficient but also pose significant safety hazards, making it difficult to meet the high requirements of modern cities for the safe operation of underground utility tunnels.

[0003] Currently, although some robot-based underground utility tunnel inspection technologies have emerged, these technologies still have many shortcomings in practical applications. For example, when existing inspection robots collect environmental monitoring data for utility tunnels, the accuracy and reliability of the data are difficult to guarantee due to the inherent precision limitations of the sensors and the complex environment of the underground utility tunnels. This leads to subsequent hazard feature extraction potentially being based on inaccurate data, thus affecting the accuracy of hazard identification.

[0004] Existing hazard feature extraction algorithms often lack effective evaluation mechanisms for extraction performance, failing to promptly identify adaptability issues in different environments, leading to unstable efficiency and accuracy in hazard identification. Furthermore, when hazard identification results are found to be poor, existing technologies lack systematic strategy optimization methods, making it difficult to quickly pinpoint the problem and implement targeted optimizations, thus impacting the overall performance of the inspection system.

[0005] Meanwhile, existing inspection systems lack a dynamic adjustment mechanism for resource allocation, failing to prioritize resources based on actual inspection needs and hazard identification, leading to resource waste and low inspection efficiency. For example, in critical areas or areas with high hazard incidence, sufficient inspection resources may not be allocated in a timely manner, thus affecting the timely detection and handling of hazards.

[0006] In addition, the existing dynamic identification system for underground utility tunnel hazards has poor coordination between its modules, and the data transmission and processing flow is not smooth enough. It cannot achieve real-time, accurate identification and dynamic monitoring of utility tunnel hazards, and it is difficult to meet the needs of intelligent and information-based management of urban underground utility tunnels. Summary of the Invention

[0007] The purpose of this invention is to provide an intelligent inspection robot for underground utility tunnels and a dynamic hazard identification system to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides an intelligent inspection robot for underground utility tunnels and a dynamic hazard identification system, the system comprising:

[0009] Environmental perception calibration module, hazard feature extraction module, and identification strategy optimization module;

[0010] The environmental perception calibration module is used to verify the perception accuracy of the pipeline environmental monitoring data collected by the inspection robot to obtain the pipeline perception index. Based on the pipeline perception index, it is determined whether to perform the hazard feature extraction. The pipeline perception index is used to quantitatively evaluate the matching degree of the pipeline environmental monitoring data collected by multiple sensors.

[0011] The hidden danger feature extraction module is used to evaluate the extraction effect of the hidden danger identification data of the utility tunnel after performing hidden danger feature extraction to obtain the utility tunnel feature index. Based on the utility tunnel feature index, it is determined whether to perform hidden danger identification strategy optimization. The utility tunnel feature index is used to comprehensively quantify the feature extraction level of the utility tunnel environmental monitoring data.

[0012] The identification strategy optimization module is used to obtain the utility tunnel optimization index after the implementation of the hazard identification strategy optimization, and to determine whether to implement the dynamic priority configuration of inspection resources based on the utility tunnel optimization index. The utility tunnel optimization index is used to comprehensively quantify the degree of compliance of the hazard identification strategy optimization.

[0013] Preferably, the specific steps for verifying the perception accuracy of the utility tunnel environmental monitoring data collected by the inspection robot to obtain the utility tunnel perception index are as follows:

[0014] The corresponding baseline data for pipe gallery monitoring is obtained by statistically verifying the environmental monitoring data of the pipe gallery collected by the inspection sensors within the preset period.

[0015] Acquire the perception and evaluation data of the utility tunnel within a preset period. The perception and evaluation data of the utility tunnel includes the total amount of perception and monitoring data of the utility tunnel, the amount of deviation data of perception and monitoring of the utility tunnel, and the correlation coefficient of perception calibration of the utility tunnel.

[0016] The utility tunnel perception index is obtained based on the baseline data of utility tunnel monitoring, the utility tunnel perception assessment data, and the relevant reference data of utility tunnel perception obtained from the preset storage repository.

[0017] The inspection sensors include infrared sensors, gas sensors, vision sensors, and temperature and humidity sensors;

[0018] The baseline data for monitoring the utility tunnel includes baseline temperature data, baseline gas concentration data, baseline image feature data, and baseline humidity data.

[0019] The reference data related to the pipe gallery sensing includes pipe gallery monitoring reference values, pipe gallery monitoring reference errors, reference sensing calibration correlation coefficients, and pipe gallery monitoring sensing correction factors.

[0020] Preferably, the specific process for determining whether to perform hazard feature extraction based on the utility tunnel perception indicators is as follows:

[0021] The perception indicators of the utility tunnel are compared with the preset recognition and perception threshold range obtained from the preset repository.

[0022] If the perception index of the utility tunnel is within the preset perception threshold range, then the hazard feature extraction is performed, and the extraction effect of the utility tunnel hazard identification data is evaluated to obtain the utility tunnel feature index.

[0023] If the perception indicators of the utility tunnel exceed the preset recognition threshold range, the extraction of hidden danger features will not be performed, and the inspection robot will be triggered to re-collect the environmental monitoring data of the utility tunnel.

[0024] Preferably, the specific steps for evaluating the extraction effect of the utility tunnel hazard identification data to obtain the utility tunnel feature indicators are as follows:

[0025] Obtain the utility tunnel feature assessment data within a preset period after extracting the characteristics of potential hazards;

[0026] The utility tunnel feature assessment data includes utility tunnel perception indicators, hazard identification feature data volume, total hazard monitoring data volume, hazard feature extraction time, hazard type matching value, and hazard feature extraction clarity.

[0027] The characteristic indicators of the utility tunnel are obtained by combining the characteristic assessment data of the utility tunnel with the characteristic reference data of the utility tunnel obtained from the preset repository.

[0028] The reference data for utility tunnel features includes the optimal utility tunnel perception index, the longest extraction time, the matching value of the hidden danger type, and the clarity of feature extraction.

[0029] Preferably, the method for obtaining the characteristic indicators of the utility tunnel is as follows:

[0030] The utility tunnel feature index is calculated by combining the matching degree between the utility tunnel perception index extracted from each hidden danger feature extraction within the preset period and the reference optimal utility tunnel perception index, the proportion of hidden danger identification feature data to the total amount of perceived hidden danger monitoring data, the difference between the hidden danger feature extraction time and the reference longest extraction time, the closeness between the hidden danger type matching value and the reference hidden danger type matching value, and the conformity between the hidden danger feature extraction clarity and the reference feature extraction clarity.

[0031] Preferably, the specific process for determining whether to implement the hazard identification strategy optimization based on the characteristic indicators of the utility tunnel is as follows:

[0032] If the characteristic indicators of the utility tunnel are within the preset characteristic threshold range, the hazard identification strategy optimization will not be implemented, and the environmental monitoring data of the utility tunnel will be preprocessed and transmitted to the management platform.

[0033] The preprocessing includes anomaly screening of utility tunnel environmental monitoring data, feature annotation of utility tunnel environmental monitoring data, and format conversion of utility tunnel environmental monitoring data.

[0034] If the characteristic indicators of the utility tunnel exceed the preset identification threshold range, the hazard identification strategy optimization will be implemented, and the optimized indicators of the utility tunnel after the hazard identification strategy optimization will be obtained.

[0035] Preferably, the specific process for optimizing the hazard identification strategy is as follows:

[0036] B1: Optimize the environmental perception parameters of the inspection robot. If the optimized pipe gallery feature indicators are within the preset identification feature threshold range, stop the hazard identification strategy optimization; otherwise, execute B2.

[0037] B2, optimize the parameters of the hidden danger feature extraction algorithm. If the optimized pipe gallery feature indicators are within the preset identification feature threshold range, stop the hidden danger identification strategy optimization; otherwise, execute B3.

[0038] B3, coordinates and optimizes the inspection path and recognition strategy of the inspection robot, and obtains the optimized indicators of the utility tunnel after the implementation of the hazard recognition strategy.

[0039] Preferably, the specific steps for obtaining the optimized utility tunnel indicators after implementing the hazard identification strategy are as follows:

[0040] Obtain the utility tunnel indicators to be optimized and the utility tunnel optimization evaluation data after the implementation of the hidden danger identification strategy;

[0041] The indicators to be optimized for the utility tunnel include the perception indicators and the characteristic indicators of the utility tunnel to be optimized.

[0042] The optimized evaluation data for the utility tunnel includes optimized sampling frequency, optimized total data volume for hazard monitoring, optimized data volume for hazard monitoring deviation, and optimized average matching value for hazard feature extraction.

[0043] The optimization indicators for the utility tunnel are obtained based on the indicators to be optimized, the optimization evaluation data of the utility tunnel, and the optimization reference data of the utility tunnel obtained from the preset storage repository.

[0044] The reference data for optimizing the utility tunnel includes the reference sampling frequency, the reference optimal utility tunnel feature index, and the reference hazard feature extraction matching value.

[0045] Preferably, the specific process for determining whether to perform dynamic priority configuration of inspection resources based on the utility tunnel optimization indicators is as follows:

[0046] If the utility tunnel optimization indicators are within the preset identification optimization range, the dynamic priority configuration of inspection resources will not be executed, and the utility tunnel characteristic indicators will be continuously monitored to see if they are within the preset identification characteristic threshold range.

[0047] If the optimization indicators of the utility tunnel exceed the preset identification and optimization range, then the dynamic priority configuration of inspection resources will be executed.

[0048] The dynamic priority configuration of inspection resources includes algorithm execution priority configuration and inspection equipment resource configuration.

[0049] Preferably, the present invention also includes an intelligent dynamic identification system for hidden dangers in underground utility tunnels, the system comprising: an environmental data acquisition and verification unit, a hidden danger feature extraction and evaluation unit, and an identification strategy optimization and judgment unit;

[0050] The environmental data acquisition and verification unit is used to verify the accuracy of the collection of environmental monitoring data of the pipe gallery collected by the inspection equipment to obtain the pipe gallery perception verification index. Based on the pipe gallery perception verification index, it is determined whether to perform the hidden danger feature extraction operation. The pipe gallery perception verification index is used to quantitatively evaluate the matching degree of pipe gallery environmental monitoring data collected by multiple types of sensors.

[0051] The hazard feature extraction and evaluation unit is used to analyze the extraction effect of the utility tunnel hazard identification data after performing the hazard feature extraction operation to obtain the utility tunnel feature evaluation index. Based on the utility tunnel feature evaluation index, it is determined whether to perform hazard identification strategy optimization. The utility tunnel feature evaluation index is used to comprehensively quantify the feature extraction level of the utility tunnel environmental monitoring data.

[0052] The identification strategy optimization judgment unit is used to obtain the utility tunnel optimization judgment index after the implementation of the hidden danger identification strategy optimization, and to determine whether to implement the dynamic priority adjustment of inspection resources based on the utility tunnel optimization judgment index. The utility tunnel optimization judgment index is used to comprehensively quantify the degree of compliance of the hidden danger identification strategy optimization.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] The intelligent inspection robot and dynamic hazard identification system for underground utility tunnels provided by this invention, through the setting of an environmental perception calibration module, can verify the perception accuracy of the environmental monitoring data of the utility tunnel collected by the inspection robot, obtain the utility tunnel perception index, and determine whether to perform hazard feature extraction based on the index. This process ensures that subsequent operations are performed only when the data perception accuracy meets the requirements, avoiding misjudgment or omission of hazards due to inaccurate data, improving data reliability, and laying a good foundation for subsequent hazard identification.

[0055] After performing feature extraction, the hazard feature extraction module evaluates the extraction results to obtain characteristic indicators for the utility tunnel. Based on these indicators, it then determines whether to optimize the hazard identification strategy. This allows for the timely detection of problems during feature extraction. When the extraction results are unsatisfactory, strategy optimization can be triggered promptly, ensuring the accuracy and effectiveness of hazard feature extraction and enabling the extracted features to more accurately reflect the actual hazard situation in the utility tunnel.

[0056] After obtaining the optimized indicators for the utility tunnel, the identification strategy optimization module determines whether to implement dynamic priority configuration of inspection resources based on these indicators. This approach allows for the rational allocation of inspection resources according to the optimized results. When the optimization of the hazard identification strategy achieves a high degree of compliance, resources are allocated appropriately, improving resource utilization efficiency and ensuring sufficient resource investment in key areas and important tasks. This, in turn, enables more efficient discovery and handling of utility tunnel hazards.

[0057] In the environmental perception calibration module, baseline data is obtained by statistically verifying the data collected by inspection sensors within a preset period. This baseline data is then combined with the utility tunnel perception assessment data and reference data to derive the utility tunnel perception index. This series of operations comprehensively verifies and evaluates the data collected by multiple sensors, quantifies the degree of data matching, and makes the sensor-collected data more accurate, reflecting the environmental conditions of the utility tunnel more realistically, thus providing reliable data support for subsequent operations.

[0058] When determining whether to extract potential hazards based on the perception indicators of the utility tunnel, extraction and evaluation are performed if the indicators are within a preset range; otherwise, data is re-collected. This mechanism ensures that subsequent feature extraction is only performed when the data accuracy meets the standards, avoiding invalid operations and improving the overall efficiency and accuracy of the system.

[0059] When evaluating the effectiveness of data extraction for identifying potential hazards in utility tunnels, a combination of various assessment and reference data is used to obtain characteristic indicators for the utility tunnels. This multi-dimensional evaluation method can comprehensively and objectively reflect the effectiveness of feature extraction, avoiding the limitations of single-dimensional evaluation, making the evaluation results more accurate and reliable, and providing a scientific basis for whether to optimize the strategy.

[0060] In the specific process of optimizing the hazard identification strategy, the environmental perception parameters, hazard feature extraction algorithm parameters, and inspection paths and identification strategies are optimized in sequence. This step-by-step and targeted optimization method can solve problems step by step, improve the efficiency and effectiveness of optimization, and enable the hazard identification strategy to better adapt to different utility tunnel environments and hazard situations.

[0061] When obtaining the optimization indicators for the utility tunnel, the various factors before and after optimization are fully considered by combining the indicators to be optimized, the optimization evaluation data, and the reference data. This ensures that the optimization indicators can accurately reflect the degree of compliance of the hazard identification strategy optimization and provide an accurate basis for the dynamic priority allocation of inspection resources.

[0062] The dynamic priority configuration of inspection resources is determined based on the optimization indicators of the utility tunnel. When the indicators exceed the range, the configuration is executed, including the algorithm execution priority and the configuration of inspection equipment resources. This dynamic adjustment mechanism can reasonably allocate resources according to the actual situation, improve the utilization efficiency of resources, ensure the efficient conduct of inspection work, promptly detect and deal with hidden dangers in the utility tunnel, and ensure the safe operation of the underground utility tunnel. Attached Figure Description

[0063] Figure 1 This is a schematic diagram illustrating the working principle of the intelligent inspection robot and dynamic hazard identification system for underground utility tunnels described in this invention.

[0064] Figure 2 A flowchart for extracting and judging potential hazards;

[0065] Figure 3 A flowchart illustrating the calculation method for characteristic indicators of utility tunnels;

[0066] Figure 4 A flowchart for optimizing the judgment of hazard identification strategies;

[0067] Figure 5 A flowchart for dynamic priority configuration of inspection resources. Detailed Implementation

[0068] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0069] Please see Figures 1-5 This invention provides an intelligent inspection robot for underground utility tunnels and a dynamic hazard identification system. The system includes: an environmental perception calibration module, a hazard feature extraction module, and an identification strategy optimization module. The specific implementation steps are as follows:

[0070] The environmental perception calibration module verifies the perception accuracy of the utility tunnel environmental monitoring data collected by the inspection robot. Specifically, it statistically verifies the utility tunnel environmental monitoring data collected by infrared sensors, gas sensors, vision sensors, and temperature and humidity sensors within a preset period, obtaining baseline data for utility tunnel monitoring, such as baseline temperature data, baseline gas concentration data, baseline image feature data, and baseline humidity data. Simultaneously, it acquires utility tunnel perception evaluation data within the preset period, including the total amount of utility tunnel perception monitoring data, the amount of utility tunnel perception monitoring deviation data, and the utility tunnel perception calibration correlation coefficient. This data is then combined with relevant reference data for utility tunnel perception obtained from a preset repository, such as reference values, reference errors, reference perception calibration correlation coefficients, and perception correction factors, to obtain a utility tunnel perception index used to quantitatively evaluate the matching degree of multi-sensor collected utility tunnel environmental monitoring data. Based on this utility tunnel perception index, it determines whether to perform hazard feature extraction. If the utility tunnel perception index is within a preset identification threshold range, hazard feature extraction is performed; otherwise, the inspection robot is triggered to re-collect utility tunnel environmental monitoring data.

[0071] After performing hazard feature extraction, the hazard feature extraction module evaluates the extraction effect of the utility tunnel hazard identification data. It acquires utility tunnel feature evaluation data within a preset period after hazard feature extraction. This data includes utility tunnel perception indicators, hazard identification feature data volume, total hazard monitoring data volume, hazard feature extraction time, hazard type matching value, and hazard feature extraction clarity. Combined with reference utility tunnel feature reference data obtained from a preset repository, such as the optimal utility tunnel perception indicators, the longest reference extraction time, the hazard type matching value, and the feature extraction clarity, the module calculates a comprehensive quantification of the utility tunnel environmental monitoring data feature extraction level based on the following: the matching degree between the utility tunnel perception indicators and the optimal utility tunnel perception indicators; the proportion of hazard identification feature data volume to the total hazard monitoring data volume; the difference between the hazard feature extraction time and the longest reference extraction time; the closeness between the hazard type matching value and the reference hazard type matching value; and the conformity between the hazard feature extraction clarity and the reference feature extraction clarity. This quantification determines whether to optimize the hazard identification strategy.

[0072] The identification strategy optimization module acquires the optimized utility tunnel indicators after implementing the optimized hazard identification strategy. These indicators are used to comprehensively quantify the degree of compliance with the optimized hazard identification strategy. It acquires the utility tunnel's perception indicators and feature indicators to be optimized, as well as optimization evaluation data such as optimized sampling frequency, optimized total data volume for perceived hazard monitoring, optimized data volume for perceived hazard monitoring deviation, and optimized average matching value for hazard feature extraction. Combined with reference sampling frequency, reference optimal utility tunnel feature indicators, and reference hazard feature extraction matching values ​​obtained from a preset repository, the module obtains the utility tunnel optimization indicators. Based on these indicators, it determines whether to implement dynamic priority configuration of inspection resources. If the optimization indicators exceed the preset identification optimization range, dynamic priority configuration of inspection resources, including algorithm execution priority configuration and inspection equipment resource configuration, is implemented. Otherwise, it is not implemented, and the module continuously monitors whether the utility tunnel feature indicators are within the preset identification feature threshold range.

[0073] Example 1:

[0074] In this embodiment, the intelligent inspection robot for underground utility tunnels verifies the perception accuracy of the collected environmental monitoring data to obtain the tunnel's perception indicators. The specific implementation is as follows: A preset period, which can be determined based on actual inspection needs and the characteristics of the tunnel environment (e.g., one day, one week, or one month), is used to statistically verify the environmental monitoring data collected by various inspection sensors. The inspection sensors used include infrared sensors, gas sensors, vision sensors, and temperature and humidity sensors. These sensors are used to collect environmental monitoring data such as temperature, gas concentration, image features, and temperature and humidity within the tunnel.

[0075] During the statistical verification process, a comprehensive inspection and verification of the data collected by the sensors is required, including aspects such as data completeness, consistency, and accuracy. For example, this involves checking for missing or outlier values ​​in the temperature data collected by the infrared sensor, ensuring the gas concentration data collected by the gas sensor is within a reasonable range, verifying the clarity and completeness of the image feature data collected by the visual sensor, and confirming that the temperature and humidity data collected by the temperature and humidity sensor matches the actual conditions of the utility tunnel environment. Through statistical verification of these data, corresponding baseline data for utility tunnel monitoring is obtained. This data includes baseline temperature data, baseline gas concentration data, baseline image feature data, and baseline humidity data. These baseline data serve as crucial evidence for subsequent verification of sensing accuracy.

[0076] Acquire utility tunnel perception assessment data within a preset period. This data includes the total amount of utility tunnel perception monitoring data, the amount of deviation data from the perception monitoring data, and the correlation coefficient for the perception calibration. The total amount of utility tunnel perception monitoring data refers to the total volume of environmental monitoring data collected by all inspection sensors within the preset period. The amount of deviation data from the perception monitoring data refers to the deviation between the collected data and the actual data, which can be determined by comparing it with known standard data or data collected by calibrated equipment. The correlation coefficient for the perception calibration reflects the degree of calibration of the sensor-collected data and the correlation between them.

[0077] When acquiring data for the assessment of utility tunnel sensing, it is crucial to ensure the accuracy and reliability of the data. For example, determining the amount of deviation data in utility tunnel sensing monitoring requires selecting appropriate standard data or calibration equipment and performing calculations according to rigorous comparison methods to avoid errors. Simultaneously, obtaining the correlation coefficients for utility tunnel sensing calibration also needs to be based on extensive historical data and experimental analysis to ensure that they accurately reflect the correlation between sensors.

[0078] The utility tunnel monitoring baseline data, utility tunnel perception assessment data, and utility tunnel perception-related reference data obtained from a pre-set repository are combined to obtain utility tunnel perception indicators. The utility tunnel perception-related reference data includes utility tunnel monitoring reference values, utility tunnel monitoring reference errors, reference perception calibration correlation coefficients, and utility tunnel monitoring perception correction factors. Utility tunnel monitoring reference values ​​are reference standards for various monitoring parameters determined based on the normal operating conditions and historical data of the utility tunnel environment; the utility tunnel monitoring reference error is the allowable deviation range between the monitoring data and the reference value; the reference perception calibration correlation coefficient is a reference value obtained through extensive experiments and verification of sensor calibration correlation; and the utility tunnel monitoring perception correction factor is used to correct the perception indicators to improve their accuracy.

[0079] When combining this data, scientific and reasonable methods are required for calculation and analysis. For example, the deviation of each monitoring parameter can be calculated by comparing the baseline data of the utility tunnel monitoring with the reference values ​​of the utility tunnel monitoring; then, by combining the amount of deviation data of the utility tunnel sensing monitoring and the reference error of the utility tunnel monitoring, it can be determined whether the deviation of the data is within the allowable range; then, the correlation coefficient between the sensors can be analyzed using the correlation coefficient between the utility tunnel sensing calibration and the reference sensing calibration; finally, the calculation results are corrected by the utility tunnel monitoring sensing correction factor to obtain the final utility tunnel sensing index.

[0080] The utility tunnel perception index is used to quantitatively evaluate the matching degree of environmental monitoring data collected by multiple sensors in the utility tunnel. This index determines whether the data collected by multiple sensors is consistent, accurate, and can truly reflect the environmental conditions of the utility tunnel. A high utility tunnel perception index indicates a good matching degree of data collected by multiple sensors and high perception accuracy; conversely, a low index indicates a poor matching degree of data and low perception accuracy, requiring further calibration or adjustment of the sensors.

[0081] The following points should be noted during the entire implementation process: First, the preset period should be reasonable, neither too long nor too short. Too long a period may lead to outdated data that cannot reflect changes in the utility tunnel environment in a timely manner, while too short a period may result in insufficient data collection, affecting the accuracy of the assessment results. Second, the statistical verification of sensor-collected data should be rigorous to ensure the accuracy of the baseline data for utility tunnel monitoring. Third, the acquisition of utility tunnel perception assessment data should be accurate to avoid inaccuracies in utility tunnel perception indicators due to data errors. Finally, when calculating based on relevant reference data for utility tunnel perception, correct methods and formulas should be used to ensure the scientific validity and rationality of the utility tunnel perception indicators.

[0082] A comprehensive data management system needs to be established to record and store the collected data, calculation processes, and results for subsequent querying and analysis. Simultaneously, the reference data related to utility tunnel sensing in the pre-set storage repository should be updated and maintained regularly to adapt to changes in the utility tunnel environment and technological advancements.

[0083] Example 2:

[0084] In this embodiment, the specific implementation method for determining whether to perform hazard feature extraction based on the pipe gallery's sensing indicators is as follows. A preset identification and sensing threshold range needs to be obtained from a preset storage library. This threshold range is pre-set based on factors such as the actual operation of the pipe gallery, sensor performance parameters, and historical data, and is used to determine whether the pipe gallery's sensing indicators meet the conditions for performing hazard feature extraction.

[0085] After obtaining the perception indicators for the utility tunnel, they are compared with a preset identification and perception threshold range. This comparison process needs to be precise and rigorous to ensure the accuracy of the data and the scientific validity of the comparison. For example, assuming the preset identification and perception threshold range is a specific numerical range, such as [X, Y], then the calculated utility tunnel perception indicators need to be compared with X and Y to determine their position within the range.

[0086] If the perception indicators of the utility tunnel are within the preset recognition threshold range, it indicates that the matching degree of the environmental monitoring data of the utility tunnel collected by multiple sensors has met the expected requirements, and the perception accuracy of the sensors can meet the needs of hazard feature extraction. In this case, the system will trigger the hazard feature extraction module to perform hazard feature extraction. Simultaneously, after performing hazard feature extraction, the extraction effect of the utility tunnel hazard identification data needs to be evaluated to obtain the utility tunnel feature indicators. During this process, it is essential to ensure the normal operation of the hazard feature extraction module, ensuring that it can accurately extract hazard-related feature data from the utility tunnel environmental monitoring data.

[0087] If the perception indicators of the utility tunnel exceed the preset recognition threshold range, whether greater than Y or less than X, it indicates that the matching degree of the data collected by the multiple sensors does not meet the requirements, and the perception accuracy of the sensors may be biased, failing to guarantee the accuracy and reliability of hazard feature extraction. In this case, the system will not perform hazard feature extraction, but will instead trigger the inspection robot to re-collect the utility tunnel environmental monitoring data. When triggering re-collection, it is necessary to clearly define the re-collection cycle, method, and sensor operating status to ensure that the newly collected data more accurately reflects the actual environmental conditions of the utility tunnel.

[0088] The role of the preset repository is crucial throughout the entire judgment process. It not only stores the preset recognition threshold range but may also contain other parameters and data related to the judgment process. Therefore, it is necessary to ensure the stability of the preset repository and the integrity of the data to avoid deviations in the judgment results due to problems with the repository.

[0089] The calculation and acquisition process of the utility tunnel sensing indicators also needs to be strictly controlled. The acquisition method of the utility tunnel sensing indicators has been described in detail in Example 1. Here, it is necessary to ensure that the utility tunnel sensing indicators used in this example have been accurately calculated and verified and have not been interfered with by data transmission, storage and other links.

[0090] In practical applications, various complex situations may be encountered. For example, the perception indicators of the utility tunnel may be right at the boundary of the threshold range. In this case, a comprehensive judgment needs to be made based on the specific situation. If necessary, other auxiliary indicators can be combined or multiple verifications can be performed to avoid misjudgment.

[0091] The system needs a robust feedback mechanism. When the inspection robot is triggered to re-collect data, this information should be promptly relayed to the relevant operators so they can understand the situation and perform appropriate checks and maintenance. For example, operators can check whether the sensors are functioning correctly, whether calibration is required, or whether there are factors in the utility tunnel environment that could affect the sensor's sensing accuracy.

[0092] To improve system reliability and stability, fault tolerance mechanisms can be implemented. For example, after triggering a re-acquisition of data, a reasonable time interval can be set to avoid affecting the normal operation of the system due to frequent re-acquisition triggers.

[0093] Throughout the implementation process, the speed of data transmission and processing must also be guaranteed. Whether it is acquiring the perception indicators of the utility tunnel or comparing them with the preset identification and perception threshold range, it must be completed within the specified time to ensure that the system can make timely judgments and responses without affecting subsequent workflows.

[0094] A comprehensive log recording system is also needed to record in detail the values ​​of the utility tunnel's sensing indicators, the comparison results with threshold ranges, whether hazard feature extraction was performed, and the time when data re-collection was triggered. This log information can not only be used for subsequent data analysis and system optimization, but also serve as a basis for tracing problems when they occur.

[0095] Example 3:

[0096] In this embodiment, the specific implementation method for evaluating the extraction effect of utility tunnel hazard identification data to obtain utility tunnel feature indicators is as follows. After performing the hazard feature extraction operation, it is necessary to obtain utility tunnel feature evaluation data within a preset period. This preset period can be determined according to the actual needs of utility tunnel inspection and the frequency of hazard identification, for example, it can be set to 1 hour, 4 hours, or 1 day. The utility tunnel feature evaluation data contains several key elements, among which the utility tunnel perception index is an index obtained by verifying the perception accuracy of sensor-collected data in the manner described in Embodiment 1, used to reflect the reliability of the current sensor data; the hazard identification feature data volume refers to the total amount of hazard-related feature data extracted from the utility tunnel environmental monitoring data within the preset period; the total hazard monitoring data volume is the total amount of all hazard monitoring-related data collected by sensors within the same preset period; the hazard feature extraction time refers to the time spent on each hazard feature extraction operation; the hazard type matching value is used to indicate the degree of matching between the extracted hazard features and the known hazard types; and the hazard feature extraction clarity reflects the clarity of the extracted hazard features.

[0097] When acquiring this data, it is essential to ensure its accuracy and completeness. For example, regarding the time required for extracting hazard features, each extraction operation should be timed using precise timing tools to avoid discrepancies in time recording. For hazard type matching values, calculations should be performed using scientific matching algorithms based on a pre-established hazard type database to ensure that the matching results accurately reflect the degree of fit between the extracted features and the actual hazard types.

[0098] Reference data on utility tunnel features is obtained from a pre-defined repository. This data includes the optimal reference utility tunnel perception index, the longest reference extraction time, the reference hazard type matching value, and the reference feature extraction clarity. The optimal reference utility tunnel perception index is an ideal perception index value determined based on historical data and the normal operating status of the utility tunnel; the longest reference extraction time is the maximum time allowed for hazard feature extraction; the reference hazard type matching value is a standard for the degree of matching between hazard features and hazard types under ideal conditions; and the reference feature extraction clarity is a reference standard for measuring the clarity of hazard feature extraction.

[0099] The characteristic indicators of the utility tunnel are obtained by comprehensively calculating multiple parameters. Specifically, it is necessary to calculate the matching degree between the tunnel perception index extracted for each hazard feature within a preset period and the reference optimal tunnel perception index. This matching degree can be reflected by the difference or ratio between the two, for example, by using (tunnel perception index / reference optimal tunnel perception index) × 100% to represent the closeness between the two; calculate the proportion of hazard identification feature data to the total amount of hazard perception monitoring data, i.e., (hazard identification feature data / total amount of hazard perception monitoring data) × 100%, which reflects the efficiency of feature extraction; calculate the difference between the hazard feature extraction time and the reference longest extraction time, and use "hazard feature extraction time - reference longest extraction time" to determine whether the extraction time is within a reasonable range; calculate the closeness between the hazard type matching value and the reference hazard type matching value, which can be done by using the absolute difference or relative difference between the two, such as |hazard type matching value - reference hazard type matching value|, the smaller the difference, the higher the closeness; calculate the conformity between the hazard feature extraction clarity and the reference feature extraction clarity, which can be determined by comparing the numerical difference between the two or by using a similarity algorithm.

[0100] The above calculation results are combined according to certain weights to obtain the characteristic index of the utility tunnel, which can be expressed by the following formula:

[0101] ;

[0102] in, This indicates the number of parameters involved in the calculation, as shown here. These correspond to the five parameters mentioned above: matching degree, proportion, difference, proximity, and conformity. For the first The weights of each parameter need to be preset according to the importance of each parameter in evaluating the feature extraction effect. For example, the weight of matching degree can be set to 0.2, the weight of proportion can be set to 0.2, the weight of difference can be set to 0.15, the weight of proximity can be set to 0.25, the weight of conformity can be set to 0.2, and the sum of all weights is 1. For the first The standardized calculation results of each parameter need to be converted into numerical values ​​with uniform dimensions through standardization processing in order to perform weighted summation.

[0103] In the formula, the utility tunnel feature index is used to comprehensively quantify the feature extraction level of the utility tunnel environmental monitoring data; its value reflects the quality of feature extraction. Weight The settings must be based on extensive experimental data and practical application experience to ensure that the importance of each parameter is reasonably reflected. Standardization processing. Methods such as min-max normalization can be used. For example, for the matching degree, assuming its value range is... The standardized result is This maps the results of each parameter to... Within the range.

[0104] During implementation, the following points should be noted: The preset period should be appropriately chosen. If the period is too short, sufficient feature extraction data may not be collected, leading to inaccurate evaluation results; if the period is too long, the current feature extraction effect cannot be reflected in a timely manner. Weight settings need to be adjusted periodically based on actual conditions. For example, when the utility tunnel environment changes or new characteristics of potential hazards emerge, the importance of each parameter needs to be reassessed, and the weights optimized. Standardization methods must be scientific and reasonable to ensure the comparability of calculation results for different parameters.

[0105] Data storage and management are also crucial. A detailed database needs to be established to record each acquired utility tunnel characteristic assessment data, intermediate results during the calculation process, and the final utility tunnel characteristic indicators for subsequent querying and analysis. Simultaneously, the accuracy and timeliness of the utility tunnel characteristic reference data in the pre-set repository must be ensured, and the reference data can be updated according to the actual operation of the utility tunnel and technological advancements.

[0106] Example 4:

[0107] In this embodiment, the specific implementation method for determining whether to implement the hazard identification strategy optimization based on the characteristic indicators of the utility tunnel is as follows.

[0108] The system will acquire the characteristic indicators of the utility tunnel calculated in the manner described in Example 3. These indicators are key values ​​for comprehensively quantifying the feature extraction level of the utility tunnel environmental monitoring data. Simultaneously, it will retrieve a preset identification feature threshold range from a preset repository. This range is a pre-set range based on various factors, including the actual needs for hazard identification during the daily operation of the utility tunnel, historical feature extraction performance data, and system performance parameters. It is used to determine whether the current feature extraction effect meets the requirements.

[0109] The system compares the utility tunnel feature indicators with the preset feature threshold range. Assuming the preset feature threshold range is [80, 100] (this is just an example; in actual applications, this range can be set according to specific circumstances), if the value of the utility tunnel feature indicator is between 80 and 100, it indicates that the current extraction effect of the utility tunnel hazard identification data has reached the system's preset standard, and the efficiency, accuracy, and clarity of feature extraction all meet the requirements of subsequent operations. In this case, the system will not perform hazard identification strategy optimization, but will instead preprocess the utility tunnel environmental monitoring data before transmitting it to the management platform.

[0110] The preprocessing process comprises three main stages: anomaly screening of utility tunnel environmental monitoring data, feature annotation of utility tunnel environmental monitoring data, and format conversion of utility tunnel environmental monitoring data. Taking temperature monitoring data within the utility tunnel as an example, in the anomaly screening stage, the system automatically identifies data exceeding the normal temperature range (such as exceeding 40℃ or falling below 0℃), marks it as anomaly data, and removes or further verifies it; in the feature annotation stage, for the extracted temperature anomaly feature data, information such as the time and location of the anomaly and the specific type of anomaly (such as localized overheating) is annotated; the format conversion stage converts temperature data collected by different sensors (which may be stored in different formats) into a standard format that the management platform can recognize, such as converting temperature data into a unified numerical format in degrees Celsius.

[0111] If the value of the utility tunnel feature index exceeds the preset feature identification threshold range, whether it is less than 80 or greater than 100 (assuming the range is [80, 100]), it indicates that the current utility tunnel hazard identification data extraction effect does not meet the system requirements. For example, when the utility tunnel feature index is 75, it may mean that the hazard feature extraction takes too long, or the hazard type matching value is low, resulting in poor overall feature extraction effect. At this time, the system will trigger the hazard identification strategy optimization process to improve the feature extraction level.

[0112] After triggering the hazard identification strategy optimization, the system will implement the optimization steps step by step according to the preset plan. First, it will optimize the environmental perception parameters, that is, adjust the perception parameters of devices such as infrared sensors and gas sensors on the inspection robot. For example, it may adjust the detection distance threshold of the infrared sensor or the sampling frequency of the gas sensor to improve the accuracy of the sensor data and thus enhance the characteristic indicators of the utility tunnel. After the adjustment is completed, the system will recalculate the characteristic indicators of the utility tunnel and determine whether they fall within the preset identification characteristic threshold range.

[0113] If, after adjusting the environmental perception parameters, the utility tunnel feature indicators still do not meet the requirements (e.g., the adjusted indicator is 78, still less than 80 (assuming the lower limit of the interval is 80), then the next step is to optimize the parameters of the hazard feature extraction algorithm. Taking image recognition algorithms as an example, parameters such as the convolution kernel size and learning rate may be adjusted to optimize the algorithm's ability to extract hazard features from utility tunnel images. After adjustment, the utility tunnel feature indicators are recalculated to determine whether the conditions are met.

[0114] If, after algorithm parameter optimization, the utility tunnel's characteristic indicators still fail to meet the standards (e.g., an indicator of 79, close to but not reaching the lower limit of the threshold range), then collaborative optimization of the inspection robot's inspection path and recognition strategy will be implemented. For example, the inspection robot's route within the utility tunnel might be changed, increasing the inspection frequency of areas prone to potential hazards, while simultaneously adjusting the recognition strategy, such as prioritizing feature extraction from images of areas with abnormal temperatures. During the collaborative optimization process, the system will acquire relevant data in real time after the optimization and calculate the utility tunnel optimization indicators to evaluate the optimization effect.

[0115] Real-time data transmission and processing are crucial throughout the entire judgment and optimization process. For example, when the characteristic indicators of the utility tunnel exceed the threshold range, the system must promptly transmit this information to the optimization module to avoid delays that could lead to untimely identification of potential hazards. Simultaneously, when the pre-processed environmental monitoring data of the utility tunnel is transmitted to the management platform, the integrity and accuracy of the data must be ensured to prevent data loss or errors during transmission.

[0116] Furthermore, the preset identification feature threshold ranges in the preset repository are not static and can be dynamically adjusted according to the actual operation of the utility tunnel. For example, when new equipment is added to the tunnel, causing environmental changes, the threshold ranges can be reset by analyzing historical data to adapt to the new operating conditions. The system also needs to have a logging function to record in detail the values ​​of each utility tunnel feature indicator, the comparison results with the threshold ranges, whether optimization was performed, and the specific steps of the optimization, so as to facilitate subsequent review and analysis and provide a reference for further optimization of the system.

[0117] Example 5:

[0118] In this embodiment, the specific implementation methods for optimizing the hazard identification strategy and determining the allocation of inspection resources based on the pipeline corridor optimization index are as follows.

[0119] When the characteristic indicators of the utility tunnel exceed the preset identification threshold range, the system initiates a hazard identification strategy optimization process, which includes three progressive optimization steps. First, step B1 is executed to optimize the environmental perception parameters of the inspection robot. Specifically, this involves adjusting various parameters of the infrared sensor, gas sensor, vision sensor, and temperature and humidity sensor, such as adjusting the temperature measurement range of the infrared sensor, the sampling period of the gas sensor, the focal length parameter of the vision sensor, or the calibration coefficient of the temperature and humidity sensor. After adjustment, the system re-collects environmental monitoring data of the utility tunnel and calculates the characteristic indicators to determine whether they fall within the preset identification threshold range. If the optimized characteristic indicators meet the requirements, optimization stops; otherwise, it proceeds to step B2.

[0120] Step B2 involves optimizing the parameters of the hazard feature extraction algorithm. Different types of algorithms correspond to different optimization parameters. For example, for deep learning-based image recognition algorithms, the number of convolutional layers, the number of neurons, the learning rate, or the number of iterations can be adjusted; for gas concentration anomaly detection algorithms, the threshold judgment parameters or the sliding window size can be optimized. After the algorithm parameters are adjusted, the system uses the new parameters to extract features from the utility tunnel environmental monitoring data and recalculates the utility tunnel feature indicators. If the indicators meet the preset range requirements, optimization stops; if they still do not meet the requirements, step B3 is executed, which involves co-optimizing the inspection path and recognition strategy of the inspection robot.

[0121] In step B3, collaborative optimization involves two aspects: firstly, adjusting the physical inspection path of the inspection robot, such as increasing the inspection frequency of high-risk areas like utility tunnel intersections and areas with concentrated equipment, or shortening the inspection interval for specific areas; secondly, optimizing the identification strategy, such as prioritizing monitoring data from areas with abnormal temperatures, or activating a multi-sensor joint detection mechanism for locations with large gas concentration fluctuations. During collaborative optimization, the system simultaneously acquires the optimized utility tunnel indicators and the utility tunnel optimization evaluation data. The optimized utility tunnel indicators include the perception indicators and feature indicators; the former reflects the matching degree of the optimized sensor data, and the latter reflects the effect of the optimized feature extraction. The utility tunnel optimization evaluation data includes the optimized sampling frequency, the total amount of optimized perception hazard monitoring data, the amount of optimized perception hazard monitoring deviation data, and the average matching value of optimized hazard feature extraction. These data are used to evaluate the actual effect of the optimization measures.

[0122] After acquiring the above data, the system combines the utility tunnel optimization indicators, the utility tunnel optimization evaluation data, and the utility tunnel optimization reference data in the preset repository to calculate the utility tunnel optimization indicators. The utility tunnel optimization reference data includes a reference sampling frequency, a reference optimal utility tunnel feature indicator, and a reference hazard feature extraction matching value. The reference sampling frequency is a sensor sampling standard set according to the normal operation requirements of the utility tunnel; the reference optimal utility tunnel feature indicator is the benchmark for feature extraction performance under ideal conditions; and the reference hazard feature extraction matching value is the ideal matching degree between features and hazard types. The utility tunnel optimization indicators are used to comprehensively quantify the degree of compliance with the hazard identification strategy optimization. The calculation process needs to comprehensively consider the weights and comparison relationships of various data points, but does not involve specific formulas; instead, it uses preset logical rules for quantitative evaluation.

[0123] Based on the utility tunnel optimization indicators, the system performs subsequent judgments: if the optimization indicators are within the preset identification and optimization range, it indicates that the current optimization measures have met the system performance requirements, and dynamic priority configuration of inspection resources is not executed. Simultaneously, the system continuously monitors whether the utility tunnel characteristic indicators remain within the preset identification feature threshold range. If the optimization indicators exceed the preset identification and optimization range, it indicates that the existing optimization measures have not yet solved the problem, and dynamic priority configuration of inspection resources needs to be executed. This configuration includes two aspects: algorithm execution priority configuration, such as setting the hazard feature extraction algorithm to high priority to ensure it occupies computing resources preferentially; and inspection equipment resource configuration, such as allocating more power resources or communication bandwidth to sensors in key areas to ensure the stability of data acquisition and transmission.

[0124] Throughout the implementation process, it is crucial to ensure that parameter adjustments for each optimization step are based on historical data and practical operational experience to avoid blind adjustments that could lead to system performance fluctuations. For example, the adjustment range of environmental perception parameters should refer to the sensor's technical manual to ensure it does not exceed its safe operating range; the optimization of algorithm parameters should be combined with the characteristics of the utility tunnel monitoring data to avoid increased feature extraction errors due to improper parameter settings. Simultaneously, the adjustment of the inspection path must consider the physical structure of the utility tunnel and the robot's mobility to ensure path feasibility and inspection efficiency.

[0125] In addition, the system needs to establish a comprehensive optimization record mechanism to record in detail the specific values ​​of each environmental perception parameter adjustment, the process of algorithm parameter optimization, the plan for changing inspection paths, and the corresponding changes in utility tunnel optimization indicators. These records are not only used to trace the optimization process, but also to provide a reference for handling similar problems in the future, forming a knowledge base for optimization strategies. The utility tunnel optimization reference data in the preset repository also needs to be updated regularly, and the reference standards should be adjusted according to changes in the operation status of the utility tunnel and technological upgrades to ensure the timeliness and accuracy of the evaluation basis.

[0126] Through the above implementation methods, the system can optimize the hazard identification process according to a progressive strategy and determine whether to adjust the inspection resource allocation based on quantitative indicators, forming a closed-loop intelligent optimization mechanism. This continuously improves the efficiency and accuracy of hazard identification in underground utility tunnels, ensuring the safe operation of the tunnels. The entire process strictly follows data-driven decision-making logic, ensuring the scientific validity and effectiveness of each step through step-by-step optimization and quantitative evaluation.

[0127] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0128] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent inspection robot for underground utility tunnels, characterized in that: include: Environmental perception calibration module, hazard feature extraction module, and identification strategy optimization module; The environmental perception calibration module is used to verify the perception accuracy of the pipeline environmental monitoring data collected by the inspection robot to obtain the pipeline perception index. Based on the pipeline perception index, it is determined whether to perform the hazard feature extraction. The pipeline perception index is used to quantitatively evaluate the matching degree of the pipeline environmental monitoring data collected by multiple sensors. The hazard feature extraction module is used to evaluate the extraction effect of the utility tunnel hazard identification data after performing hazard feature extraction to obtain utility tunnel feature indicators. Based on the utility tunnel feature indicators, it is determined whether to perform hazard identification strategy optimization. The utility tunnel feature indicators are used to comprehensively quantify the feature extraction level of utility tunnel environmental monitoring data. The identification strategy optimization module is used to obtain the utility tunnel optimization index after the implementation of the hazard identification strategy optimization, and to determine whether to implement the dynamic priority configuration of inspection resources based on the utility tunnel optimization index. The utility tunnel optimization index is used to comprehensively quantify the degree of compliance of the hazard identification strategy optimization. The specific steps for verifying the perception accuracy of the pipeline corridor environmental monitoring data collected by the inspection robot to obtain the pipeline corridor perception index are as follows: The corresponding baseline data for pipe gallery monitoring is obtained by statistically verifying the environmental monitoring data of the pipe gallery collected by the inspection sensors within the preset period. Acquire the perception and evaluation data of the utility tunnel within a preset period. The perception and evaluation data of the utility tunnel includes the total amount of perception and monitoring data of the utility tunnel, the amount of deviation data of perception and monitoring of the utility tunnel, and the correlation coefficient of perception calibration of the utility tunnel. The utility tunnel perception index is obtained based on the baseline data of utility tunnel monitoring, the utility tunnel perception assessment data, and the relevant reference data of utility tunnel perception obtained from the preset storage repository. The inspection sensors include infrared sensors, gas sensors, vision sensors, and temperature and humidity sensors; The baseline data for monitoring the utility tunnel includes baseline temperature data, baseline gas concentration data, baseline image feature data, and baseline humidity data. The reference data related to the pipe gallery sensing includes pipe gallery monitoring reference values, pipe gallery monitoring reference errors, reference sensing calibration correlation coefficients, and pipe gallery monitoring sensing correction factors.

2. The intelligent inspection robot for underground utility tunnels as described in claim 1, characterized in that, The specific process for determining whether to perform hazard feature extraction based on utility tunnel perception indicators is as follows: The perception indicators of the utility tunnel are compared with the preset recognition and perception threshold range obtained from the preset repository. If the perception index of the utility tunnel is within the preset perception threshold range, then the hazard feature extraction is performed, and the extraction effect of the utility tunnel hazard identification data is evaluated to obtain the utility tunnel feature index. If the perception indicators of the utility tunnel exceed the preset recognition threshold range, the extraction of hidden danger features will not be performed, and the inspection robot will be triggered to re-collect the environmental monitoring data of the utility tunnel.

3. The intelligent inspection robot for underground utility tunnels as described in claim 2, characterized in that, The specific steps for evaluating the extraction effect of utility tunnel hazard identification data to obtain utility tunnel feature indicators are as follows: Obtain the utility tunnel feature assessment data within a preset period after extracting the characteristics of potential hazards; The utility tunnel feature assessment data includes utility tunnel perception indicators, hazard identification feature data volume, total hazard monitoring data volume, hazard feature extraction time, hazard type matching value, and hazard feature extraction clarity. The characteristic indicators of the utility tunnel are obtained by combining the characteristic assessment data of the utility tunnel with the characteristic reference data of the utility tunnel obtained from the preset repository. The reference data for utility tunnel features includes the optimal utility tunnel perception index, the longest extraction time, the matching value of the hidden danger type, and the clarity of feature extraction.

4. The intelligent inspection robot for underground utility tunnels as described in claim 3, characterized in that, The method for obtaining the characteristic indicators of the utility tunnel is as follows: The utility tunnel feature index is calculated by combining the matching degree between the utility tunnel perception index extracted from each hidden danger feature extraction within the preset period and the reference optimal utility tunnel perception index, the proportion of hidden danger identification feature data to the total amount of perceived hidden danger monitoring data, the difference between the hidden danger feature extraction time and the reference longest extraction time, the closeness between the hidden danger type matching value and the reference hidden danger type matching value, and the conformity between the hidden danger feature extraction clarity and the reference feature extraction clarity.

5. The intelligent inspection robot for underground utility tunnels as described in claim 1, characterized in that, The specific process for determining whether to implement the hazard identification strategy optimization based on the characteristic indicators of the utility tunnel is as follows: If the characteristic indicators of the utility tunnel are within the preset characteristic threshold range, the hazard identification strategy optimization will not be implemented, and the environmental monitoring data of the utility tunnel will be preprocessed and transmitted to the management platform. The preprocessing includes anomaly screening of utility tunnel environmental monitoring data, feature annotation of utility tunnel environmental monitoring data, and format conversion of utility tunnel environmental monitoring data. If the characteristic indicators of the utility tunnel exceed the preset identification threshold range, the hazard identification strategy optimization will be implemented, and the optimized indicators of the utility tunnel after the hazard identification strategy optimization will be obtained.

6. The intelligent inspection robot for underground utility tunnels as described in claim 5, characterized in that, The specific process for optimizing the hazard identification strategy is as follows: B1: Optimize the environmental perception parameters of the inspection robot. If the optimized pipe gallery feature indicators are within the preset identification feature threshold range, stop the hazard identification strategy optimization; otherwise, execute B2. B2, optimize the parameters of the hidden danger feature extraction algorithm. If the optimized pipe gallery feature indicators are within the preset identification feature threshold range, stop the hidden danger identification strategy optimization; otherwise, execute B3. B3, coordinates and optimizes the inspection path and recognition strategy of the inspection robot, and obtains the optimized indicators of the utility tunnel after the implementation of the hazard recognition strategy.

7. The intelligent inspection robot for underground utility tunnels as described in claim 5, characterized in that, The specific steps for obtaining the optimized utility tunnel indicators after implementing the hazard identification strategy are as follows: Obtain the utility tunnel indicators to be optimized and the utility tunnel optimization evaluation data after the implementation of the hidden danger identification strategy; The indicators to be optimized for the utility tunnel include the perception indicators and the characteristic indicators of the utility tunnel to be optimized. The optimized evaluation data for the utility tunnel includes optimized sampling frequency, optimized total data volume for hazard monitoring, optimized data volume for hazard monitoring deviation, and optimized average matching value for hazard feature extraction. The optimization indicators for the utility tunnel are obtained based on the indicators to be optimized, the optimization evaluation data of the utility tunnel, and the optimization reference data of the utility tunnel obtained from the preset storage repository. The reference data for optimizing the utility tunnel includes the reference sampling frequency, the reference optimal utility tunnel feature index, and the reference hazard feature extraction matching value.

8. The intelligent inspection robot for underground utility tunnels as described in claim 1, characterized in that, The specific process for determining whether to perform dynamic priority configuration of inspection resources based on utility tunnel optimization indicators is as follows: If the utility tunnel optimization indicators are within the preset identification optimization range, the dynamic priority configuration of inspection resources will not be executed, and the utility tunnel characteristic indicators will be continuously monitored to see if they are within the preset identification characteristic threshold range. If the optimization indicators of the utility tunnel exceed the preset identification and optimization range, then the dynamic priority configuration of inspection resources will be executed. The dynamic priority configuration of inspection resources includes algorithm execution priority configuration and inspection equipment resource configuration.

9. An intelligent dynamic identification system for hidden dangers in underground utility tunnels, applied to the intelligent inspection robot for underground utility tunnels as described in any one of claims 1 to 8, characterized in that, include: Environmental data acquisition and verification unit, hazard feature extraction and assessment unit, and identification strategy optimization and judgment unit; The environmental data acquisition and verification unit is used to verify the accuracy of the collection of environmental monitoring data of the pipe gallery collected by the inspection equipment to obtain the pipe gallery perception verification index. Based on the pipe gallery perception verification index, it is determined whether to perform the hidden danger feature extraction operation. The pipe gallery perception verification index is used to quantitatively evaluate the matching degree of pipe gallery environmental monitoring data collected by multiple types of sensors. The hazard feature extraction and evaluation unit is used to analyze the extraction effect of the utility tunnel hazard identification data after performing the hazard feature extraction operation to obtain the utility tunnel feature evaluation index. Based on the utility tunnel feature evaluation index, it is determined whether to perform hazard identification strategy optimization. The utility tunnel feature evaluation index is used to comprehensively quantify the feature extraction level of the utility tunnel environmental monitoring data. The identification strategy optimization judgment unit is used to obtain the utility tunnel optimization judgment index after the implementation of the hidden danger identification strategy optimization, and to determine whether to implement the dynamic priority adjustment of inspection resources based on the utility tunnel optimization judgment index. The utility tunnel optimization judgment index is used to comprehensively quantify the degree of compliance of the hidden danger identification strategy optimization.

Images