Underground space infrastructure state monitoring management system based on Internet of Things

Abstract

The invention relates to the technical field of state monitoring, in particular to an underground space infrastructure state monitoring management system based on the Internet of Things, which comprises a data acquisition marking module, a dynamic threshold adjustment module, a pattern analysis learning module, an anomaly identification prediction module, a task sorting scheduling module and a maintenance task management module. According to the method, through dynamic adjustment of the moving average and the exponentially weighted moving average and refining processing of data points, it is ensured that tiny variation is recognized in a changeable underground environment, so that the accuracy of fault prediction is greatly improved, and the fault prediction accuracy is improved by utilizing the long and short-term memory network and the sequence-to-sequence model. The system can automatically learn and identify the abnormal mode from complex multi-dimensional data, the response capability of early warning and the early warning accuracy are enhanced, the maintenance response process is optimized through intelligent task sorting and resource allocation, the time delay for coping with an emergency condition is effectively reduced, and the operation and maintenance efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of condition monitoring, and particularly to an underground space infrastructure condition monitoring and management system based on the Internet of Things. Background Art

[0002] Condition monitoring technology mainly involves continuously or periodically detecting and analyzing the operating conditions of machines, equipment, structures or systems, with the aim of identifying potential faults and maintenance requirements in advance to prevent unexpected shutdowns and extend the equipment life. It usually includes using sensors and monitoring devices to collect data on temperature, pressure, vibration, sound and other key performance indicators. By analyzing the data, abnormal patterns can be detected in a timely manner, so as to intervene before the problem becomes a significant fault.

[0003] Among them, the underground space infrastructure condition monitoring and management system is a technical system specifically designed to monitor underground facilities such as tunnels, subway systems, water supply and drainage systems, etc. Its main purpose is to monitor the structural integrity and operating conditions of underground structures in real time, so as to detect and handle problems such as cracks, water seepage, structural displacement in a timely manner. This can not only reduce the risk of safety accidents and improve public safety, but also help relevant management agencies optimize the timing and resource allocation of maintenance and repair operations, thereby reducing maintenance costs and extending the service life of the infrastructure.

[0004] Most of the existing technologies make abnormal judgments based on fixed thresholds, which are difficult to adapt to environmental changes, and are prone to false alarms or missed alarms under the dynamically changing underground space conditions. Lack of data analysis models makes it difficult for existing technologies to extract useful information from massive data, and the processing effect on non-linear and multi-dimensional data is particularly limited, resulting in the inability to predict and diagnose potential structural problems in real time. In addition, task scheduling and resource allocation are not flexible, unable to respond immediately to the rapid changes in the underground environment, resulting in low maintenance efficiency, increased operating costs, and failure to maximize the life of the facilities, increasing the risk of faults and safety accidents. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in the prior art, and to propose an underground space infrastructure condition monitoring and management system based on the Internet of Things.

[0006] To achieve the above purpose, the present invention adopts the following technical solution: An underground space infrastructure condition monitoring and management system based on the Internet of Things includes: The data acquisition and marking module, based on Internet of Things technology, acquires multi-dimensional data of underground space infrastructure, preliminarily screens the data and marks time stamps, and extracts key time points in the data to generate a time series data set; The dynamic threshold adjustment module adjusts the moving average and exponentially weighted moving average of historical data based on the time series dataset, performs data point matching, sets dynamic thresholds for groundwater level changes, and judges the abnormality of current data to generate abnormality recognition metrics; The pattern analysis and learning module, based on the abnormality recognition metrics, uses the LSTM model to learn normal and abnormal patterns from the multi-dimensional data of the underground space infrastructure, and analyzes the current data according to the patterns to generate a pattern recognition overview; The abnormality recognition and prediction module extracts key features from the pattern learning results, uses Seq2Seq to identify abnormal patterns, and predicts future potential problems and risks based on the identified features to generate a risk prediction summary; The task ranking and scheduling module, based on the pattern recognition overview and the risk prediction summary, uses linear programming and integer programming to rank and schedule maintenance and inspection tasks, and dynamically adjusts task priorities according to the current and predicted situations to generate a scheduling optimization path; The maintenance task management module, based on the scheduling optimization path, dynamically allocates maintenance tasks according to the set priorities and real-time monitoring data, verifies the rationality of resource allocation in key underground areas, monitors the task execution process and adjusts the response process to generate a resource allocation analysis result.

[0007] As a further solution of the present invention, the time series dataset includes data points, time tags, and key time intervals. The abnormality recognition metrics include threshold levels, abnormality scores, and trend deviation information. The pattern recognition overview includes recognized patterns, abnormality degrees, and normal behavior indicators. The risk prediction summary includes risk types, impact probabilities, and warning levels. The scheduling optimization path includes task numbers, priority levels, and estimated processing durations.

[0008] As a further solution of the present invention, the data acquisition and marking module includes: The data pre-screening sub-module, based on Internet of Things technology, collects multi-dimensional data of the underground space infrastructure, screens the data with signal strengths that meet the requirements, and eliminates data points with weak or incorrect signals by evaluating signal quality to generate a set of valid data; The timestamp marking sub-module, based on the set of valid data, assigns corresponding reception times to each piece of data, and verifies the time information attached to each piece of data through time synchronization technology to generate a time index set; The key time point extraction sub-module, based on the time index set, analyzes data fluctuation points and marks the key time points of data changes, including equipment failure or abnormal operation times, to generate a time series dataset.

[0009] As a further solution of the present invention, the dynamic threshold adjustment module includes: The historical data analysis sub-module calculates data statistical metrics based on the time series dataset, including moving average and exponentially weighted moving average, smooths the fluctuations of the time series data, and generates a smoothed time series; The threshold setting sub-module dynamically adjusts the threshold based on the smoothed time series, historical data, and the current environment, sets the data threshold matching the current conditions, and generates a threshold adjustment parameter; The anomaly judgment sub-module monitors the current data based on the threshold adjustment parameter, identifies abnormal or out-of-threshold data points, marks and records them, and generates an anomaly identification metric.

[0010] As a further solution of the present invention, the pattern analysis and learning module includes: The pattern learning sub-module analyzes the multi-dimensional data of the underground space infrastructure using the LSTM model based on the anomaly identification metric, identifies the differences in behavior patterns in the time series, classifies the normal and abnormal behaviors in the dataset, and generates a behavior pattern classification; The pattern analysis sub-module performs a matching analysis of the data patterns based on the behavior pattern classification, compares the current data with the learned patterns, identifies the pattern deviations in the data, and generates pattern deviation identification information; The overview generation sub-module organizes the data pattern information based on the pattern deviation identification information, including the identification information of the data pattern and the behavior analysis results, and analyzes based on the current data to obtain a pattern recognition overview.

[0011] As a further solution of the present invention, the anomaly identification and prediction module includes: The feature extraction sub-module extracts key features from the pattern learning results, uses Seq2Seq to analyze the behavior features of abnormal and normal data, performs key feature screening, identifies and extracts the key variables affecting the pattern judgment, and generates a set of key variables; The pattern recognition sub-module performs a sequence analysis of the data based on the set of key variables, identifies the abnormal behavior patterns in the data through behavior pattern matching, classifies and records the abnormal behaviors, and generates abnormal behavior identification information; The risk prediction sub-module performs a trend analysis of the identified abnormal behavior patterns and evaluates the potential risks, predicts the future potential risk development direction, and generates a risk prediction summary.

[0012] As a further solution of the present invention, according to the Seq2Seq, the variables affecting the pattern judgment are identified and extracted from the abnormal and normal data behaviors, according to the formula: ; Calculate the feature score of each data point , where, represents the feature weight, represents the corresponding feature value, represents the number of features.

[0013] As a further solution of the present invention, the task scheduling module includes: The task priority adjustment sub-module analyzes the risk level and urgency of multi-tasks based on the pattern recognition overview and risk prediction summary, adjusts the priorities of maintenance and inspection tasks, updates the priorities of high-risk areas in real time, and generates a priority update list; The linear programming sub-module calculates the optimal resource allocation and time management process based on the priority update list, adjusts the task sequence to match the current available resources and time window through the linear programming algorithm, optimizes the task execution order, and generates an optimal task sequence; The integer programming sub-module performs task scheduling based on the optimal task sequence through the integer programming algorithm, combines resource limitations and task divisibility, executes the optimal path of the task, and generates a scheduling optimization path.

[0014] As a further solution of the present invention, according to the linear programming algorithm, calculate the optimal resource allocation and time management process, match the current available resources and time window, according to the formula: ; Calculate the minimization of resource cost. In the formula, represents the total cost, represents the unit price of the th item of resources, represents the quantity of resources allocated to the th item of task,

[0015] As a further solution of the present invention, the maintenance task management module includes: The task allocation sub-module dynamically configures maintenance personnel and equipment to the critical operation area based on the scheduling optimization path, combines real-time monitoring data and priorities, optimizes maintenance efficiency and response speed, and generates a maintenance task configuration; The resource verification sub-module analyzes the resource usage and allocation efficiency in real time based on the maintenance task configuration, verifies the rationality of the resource configuration in the critical underground area, and generates resource rationality evaluation information; The task monitoring sub-module monitors the implementation of maintenance tasks based on the resource rationality evaluation information, adjusts the operation process, matches the on-site situation and emergencies, and optimizes the task response and completion time, and generates a resource allocation analysis result.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, through the dynamic adjustment of moving average and exponentially weighted moving average, data points are refined to ensure the identification of minor variations in the changing underground environment, thereby significantly improving the accuracy of fault prediction. By using long short-term memory networks and sequence-to-sequence models, the system can automatically learn and identify abnormal patterns from complex multi-dimensional data, enhancing the response ability and early warning accuracy of early alerts. Through intelligent task sequencing and resource allocation, the maintenance response process is optimized, effectively reducing the time delay in responding to emergencies, not only improving the operation and maintenance efficiency, but also providing data-driven decision support for management, thus achieving the maximization of cost-effectiveness and the optimization of facility operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is the system flow chart of the present invention; Figure 2 is the schematic diagram of the system framework of the present invention; Figure 3 is the flow chart of the data acquisition and marking module of the present invention; Figure 4 is the flow chart of the dynamic threshold adjustment module of the present invention; Figure 5 is the flow chart of the pattern analysis and learning module of the present invention; Figure 6 is the flow chart of the abnormal identification and prediction module of the present invention; Figure 7 is the flow chart of the task sequencing and scheduling module of the present invention; Figure 8 is the flow chart of the maintenance task management module of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention.

[0019] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as limiting the present invention. In addition, in the description of the present invention, the meaning of "a plurality" is two or more unless otherwise specifically defined.

[0020] Embodiment 1 Please refer to Figure 1 , the present invention provides a technical solution: an underground space infrastructure status monitoring and management system based on the Internet of Things includes: The data acquisition and marking module, based on Internet of Things technology, acquires multi-dimensional data of underground space infrastructure, preliminarily screens the data and marks time stamps, extracts key time points from the data, and generates a time series data set; the underground space infrastructure specifically refers to underground pipelines, integrated pipe galleries, underground passages, underground public parking lots, civil air defense facilities, rail transit, etc.; the Internet of Things technology specifically refers to continuously collecting multi-dimensional data through sensors at each key point of underground facilities, and sending the collected data to a processing system or cloud platform through wireless or wired networks.

[0021] The dynamic threshold adjustment module, based on the time series data set, adjusts the moving average and exponentially weighted moving average of historical data, performs data point matching, sets a dynamic threshold for underground water level changes, and judges the abnormality of current data to generate an anomaly recognition index; The pattern analysis and learning module, based on the anomaly recognition index, uses the LSTM model to learn normal and abnormal patterns from the multi-dimensional data of underground space infrastructure, and analyzes the current data according to the patterns to generate a pattern recognition overview; The anomaly recognition and prediction module extracts key features from the pattern learning results, uses Seq2Seq to identify abnormal patterns, and predicts future potential problems and risks according to the identified features to generate a risk prediction summary; The task ranking and scheduling module, based on the pattern recognition overview and risk prediction summary, uses linear programming and integer programming to rank and schedule maintenance and inspection tasks, and dynamically adjusts task priorities according to current and predicted situations to generate a scheduling optimization path; The maintenance task management module, based on the scheduling optimization path, dynamically allocates maintenance tasks according to the set priorities and real-time monitoring data, verifies the rationality of resource allocation in key underground areas, monitors the task execution process and adjusts the response process to generate a resource allocation analysis result.

[0022] The time series data set includes data points, time tags, and key time intervals. The anomaly recognition index includes threshold levels, anomaly scores, and trend deviation information. The pattern recognition overview includes recognized patterns, anomaly degrees, and normal behavior indicators. The risk prediction summary includes risk types, impact probabilities, and warning levels. The scheduling optimization path includes task numbers, priority levels, and estimated processing durations.

[0023] Please refer to Figure 2 and Figure 3 , the data acquisition and marking module includes a data preliminary screening sub-module, a time stamp marking sub-module, and a key time point extraction sub-module; The data pre-screening sub-module, based on Internet of Things technology, collects multi-dimensional data of underground space infrastructure, screens out valid data with conforming signal strength, evaluates signal quality, eliminates weak or incorrect data points, and generates a set of valid data; Based on the received multi-dimensional data, a preliminary assessment of signal strength is carried out. Data with signal strength lower than the set threshold is directly excluded. For data that meets the threshold, a signal quality assessment algorithm, such as the calculation of signal-to-noise ratio (SNR), is further used to determine the reliability of the data. The signal quality assessment process involves measuring the signal-to-noise ratio of data points and comparing the signal-to-noise ratio of each data point with a preset minimum acceptable value. A data point is considered valid only when its signal-to-noise ratio is higher than this threshold. In addition, any incorrect data points caused by signal interference or transmission errors will also be identified and eliminated through error detection and correction (EDAC), ensuring that only high-quality data can enter the next stage of processing to generate a set of valid data.

[0024] The timestamp marking sub-module, based on the set of valid data, assigns the corresponding reception time to each piece of data, verifies the time information attached to each piece of data through time synchronization technology, and generates a set of time indexes; After obtaining the set of valid data, a specific reception timestamp is assigned to each piece of data, including time synchronization of the received data signal, using the Network Time Protocol (NTP) or clock synchronization technology to ensure the accuracy of time information. Each piece of data will be checked for the rationality of its timestamp and the deviation from the system time within the allowed error range through a time verification algorithm. Data with inaccurate or abnormal timestamps will be marked and further reviewed. Precise time marking enables the data to be sorted and indexed in chronological order, generating a set of time indexes.

[0025] The key time point extraction sub-module, based on the set of time indexes, analyzes data fluctuation points, marks the key time points of data changes, including equipment failure or abnormal operation time, and generates a time series data set; Using the set of time indexes to identify key change points in the data involves detailed fluctuation analysis of the data, using moving average or standard deviation analysis to identify when data points deviate significantly from their typical behavior. In addition, time points related to equipment failure or abnormal operation are automatically marked according to a rule set. For example, when temperature or pressure exceeds the safety threshold, all key time points will be integrated into the time series data set, providing a basis for fault diagnosis and maintenance decision-making.

[0026] Please refer to Figure 2 and Figure 4 The dynamic threshold adjustment module includes a historical data analysis sub-module, a threshold setting sub-module, and an anomaly judgment sub-module; The historical data analysis sub-module calculates data statistical metrics based on a time series dataset, including moving average and exponentially weighted moving average, to smooth the fluctuations in the time series data and generate a smoothed time series; Smoothing is performed through the calculation methods of moving average and exponentially weighted moving average, including calculating the surrounding data of each data point to smooth short-term fluctuations. Specifically, the average of the past N data points is used as the moving average of the current point, and at the same time, higher weights are applied to the most recent data points to calculate the exponentially weighted moving average. The weights are automatically adjusted according to the relative importance of the past data points and the new data points. As a result, the fluctuations in the time series data are effectively suppressed, thereby reducing the misunderstanding caused by random fluctuations. The processed data can more truly reflect the trend changes, providing a more stable data basis for the next threshold setting and generating a smoothed time series.

[0027] The threshold setting sub-module dynamically adjusts the threshold based on the smoothed time series, sets the data threshold that matches the current conditions, and generates threshold adjustment parameters; Receives the smoothed time series data and dynamically adjusts the threshold according to historical data and the current environment, including comparing the statistical characteristics of the historical time series with the real-time characteristics of the current data to determine the data threshold that best suits the current situation, and dynamically adjusts the threshold to ensure that the data threshold is updated in real time as the environment changes. The adjustment is set by analyzing the fluctuation patterns and their frequencies in the historical data, ensuring that the threshold can capture important data changes and avoid frequent false alarms caused by over-sensitivity, and generates threshold adjustment parameters.

[0028] The anomaly judgment sub-module monitors the current data based on the threshold adjustment parameters, identifies abnormal or out-of-threshold data points, and marks and records them to generate anomaly identification metrics; Utilizes the threshold adjustment parameters to monitor the current data in real time, identifies abnormal or out-of-threshold data points, which involves an immediate comparison of the data points with the threshold. Once a data point exceeds the threshold, the system will automatically mark and record it, including recording the specific time and value of the data point for further analysis. In addition, the marked abnormal data points will be further processed to determine whether they are real anomalies or accidental errors, ensuring that only real anomalies are recorded and reported, generating anomaly identification metrics, and providing key data anomaly information.

[0029] Please refer to Figure 2 and Figure 5 , the pattern analysis and learning module includes a pattern learning sub-module, a pattern analysis sub-module, and an overview generation sub-module; Based on the anomaly recognition metrics, the pattern learning sub-module uses the LSTM model to analyze the multi-dimensional data of the underground space infrastructure, identify the differences in behavior patterns in the time series, classify the normal and abnormal behaviors in the data set, and generate the behavior pattern classification. According to the anomaly recognition metrics, the LSTM model is applied to analyze the multi-dimensional data to identify the differences in behavior patterns in the time series. This involves deep learning on the historical sequence of data points, using long short-term memory networks to capture the long-term dependencies and pattern changes in the data. The LSTM network effectively distinguishes the important features and their time points in the data through its gating mechanism, thereby identifying the normal and abnormal behavior patterns. Each data point is automatically classified as normal or abnormal according to its behavior characteristics. This classification is not only based on a single data point, but also takes into account the context information in the continuous sequence, thus improving the accuracy and reliability of the classification and generating the behavior pattern classification result.

[0030] Based on the behavior pattern classification, the pattern analysis sub-module conducts a matching analysis of the data patterns, compares the current data with the learned patterns, identifies the pattern deviations in the data, and generates the pattern deviation identification information. According to the behavior pattern classification, a matching analysis of the data patterns is performed. In this process, the current data is carefully compared with the learned patterns. By analyzing the pattern deviations in the data, including comparing the consistency between the data behavior at each time point and the expected pattern, the deviation analysis uses the established classification model to evaluate the differences between the current behavior and the historical behavior, identifies the patterns that are significantly different from the normal behavior, and helps to detect potential equipment failures or operation anomalies at an early stage, generating the pattern deviation identification information.

[0031] Based on the pattern deviation identification information, the overview generation sub-module sorts out the data pattern information, including the identification information of the data patterns and the behavior analysis results, and analyzes according to the current data to obtain the pattern recognition overview. By analyzing the collected pattern deviation data, including the occurrence frequency, correlation of each data pattern and its impact on the system operation, a comprehensive report containing the identification information of the data patterns and the behavior analysis results is sorted out, which records the characteristics of various patterns and abnormal behaviors, provides a view of the data patterns and system behaviors. Based on the analysis of the current data and historical behavior patterns, the overview helps maintenance personnel better understand the facility operation status and potential problems.

[0032] Please refer to Figure 2 and Figure 6 For the anomaly recognition and prediction module, it includes a feature extraction sub-module, a pattern recognition sub-module, and a risk prediction sub-module. The feature extraction sub-module extracts key features from the pattern learning results, uses Seq2Seq to analyze the behavioral features of abnormal and normal data, conducts key feature screening, identifies and extracts the key variables affecting pattern judgment, and generates a set of key variables; Through Seq2Seq, variables affecting pattern judgment are identified and extracted from abnormal and normal data behaviors, according to the formula: ; Calculate the feature score of each data point for judging its behavior pattern. In the formula, represents the feature weight, represents the corresponding feature value, represents the number of features; Let , indicating that four key features are involved in the analysis. Among them, and , by analyzing the historical failure data of different devices, determine the statistical significance of each feature in failure prediction. Considering a data point, where (mean), (variance), (kurtosis) and (skewness), substitute into the formula for calculation: ; ; The result shows that if the abnormal standard threshold is 10, greater than 10 is abnormal, less than 10 is normal, then the calculated feature score is abnormal and is used to affect the behavior pattern judgment.

[0033] Based on the set of key variables, the pattern recognition sub-module conducts sequence analysis on the data, identifies the abnormal behavior patterns in the data through behavior pattern matching, classifies and records the abnormal behaviors, and generates abnormal behavior recognition information; After receiving the set of key variables, conduct sequence analysis on the data, including using machine learning algorithms to identify and classify the behavior patterns of each data point in the sequence data, evaluate the normal or abnormal state of the data point based on the weights and combinations of the key variables, and through comparing with the known behavior pattern library, can effectively identify the abnormal behavior patterns in the data, classify and record the behavior patterns, generate abnormal behavior recognition information, which records various abnormal states and their occurrence contexts.

[0034] Based on the abnormal behavior recognition information, the risk prediction sub-module conducts trend analysis on the identified abnormal behavior patterns and evaluates the potential risks, predicts the future potential risk development direction, and generates a risk prediction summary; Based on the abnormal behavior recognition information, perform trend analysis and evaluate potential risks. Use regression analysis and time series prediction to evaluate the development trend and potential impact of behaviors. By calculating the risk development directions in various future scenarios, the evaluation results include risk levels and occurrence probabilities, generating a risk prediction summary, providing a comprehensive view and early warning information on potential risk events, and helping to make early decisions to take measures to mitigate or avoid the impacts of potential risks.

[0035] Please refer to Figure 2 and Figure 7 , the task scheduling module includes a task priority adjustment sub-module, a linear programming sub-module, and an integer programming sub-module; The task priority adjustment sub-module analyzes the risk levels and urgencies of multiple tasks based on the pattern recognition overview and the risk prediction summary, adjusts the priorities of maintenance and inspection tasks, updates the priorities of high-risk areas in real time, and generates a priority update list; Receiving the pattern recognition overview and the risk prediction summary, analyzing the risk levels and urgencies of each task. This process involves comparing the risk scores and time sensitivities of different tasks. By identifying high-risk and high-urgency tasks, dynamically adjusting the task priorities, giving priority to tasks that affect system stability or security, updating the task priorities in real time, especially when high-risk areas are identified, ensuring that critical maintenance and inspection tasks are executed first. This process outputs a priority update list, and the list contains the new priorities of all tasks.

[0036] The linear programming sub-module calculates the optimal resource allocation and time management processes based on the priority update list, adjusts the task sequence to match the current available resources and time windows, optimizes the task execution order, and generates an optimal task sequence; Through the linear programming algorithm, calculate the optimal resource allocation and time management processes, match the current available resources and time windows, according to the formula: ; Calculate the minimization of resource costs, where represents the total cost, represents the unit price of the th item of resources, represents the quantity of resources allocated to the th task, represents the number of tasks.

[0037] If there is the following data, which are manpower, equipment, and time respectively, the manpower cost , the equipment cost , the time window cost , and there is a specific task configuration requirement, the manpower required is , and the equipment required is , time window limit , substitute into the formula and calculate to get: ; ; This result indicates that, based on the current resource configuration and cost parameters, the total cost of executing this task configuration is 1400, which helps to determine the feasibility of the task and priority adjustment, ensure the efficiency of resource use in terms of maximizing cost - effectiveness, and at the same time meet the time and resource requirements of the task.

[0038] The integer programming sub - module performs task scheduling based on the optimal task sequence through the integer programming algorithm. Combining resource constraints and task divisibility, it executes the optimal task path and generates a scheduling optimization path; According to the optimal task sequence, use the integer programming algorithm for refined task scheduling. In this process, consider the indivisibility of tasks and resource constraints, such as the availability of personnel, equipment, and time. By setting integer decision variables and related integer programming models, optimize the execution path of tasks to ensure that each task is carried out at the most appropriate time and with the least amount of resources used, obtaining a scheduling optimization path. The path plans the execution order of each task and resource allocation, providing specific operation directions for execution management.

[0039] Please refer to Figure 2 and Figure 8 , the maintenance task management module includes a task assignment sub - module, a resource verification sub - module, and a task monitoring sub - module; The task assignment sub - module, based on the scheduling optimization path, combines real - time monitoring data and priorities to dynamically configure maintenance personnel and equipment to critical operation areas, optimize maintenance efficiency and response speed, and generate a maintenance task configuration; Use the scheduling optimization path and real - time monitoring data to dynamically configure maintenance personnel and equipment to critical operation areas, including analyzing the geographical location, urgency, and resource requirements of each task, and then deciding on resource allocation based on priorities and resource availability. By optimizing maintenance efficiency and response speed, ensure rapid response in critical areas, and continuously adjust resource configuration through real - time data to match changes in task requirements, generating a maintenance task configuration, including personnel and equipment configuration for each task, for the next - step efficiency and rationality analysis.

[0040] The resource verification sub - module, based on the maintenance task configuration, analyzes the efficiency of resource use and allocation in real - time, verifies the rationality of resource configuration in critical underground areas, and generates resource rationality evaluation information; According to the maintenance task configuration, the usage and allocation efficiency of resources are analyzed in real time. This process involves checking whether the resources allocated to each task meet the predetermined efficiency standards and maintenance requirements, evaluating the applicability and effectiveness of resource allocation through simulation and comparison with historical data, ensuring that the resource allocation for each task can achieve the best efficiency and response speed, generating resource rationality evaluation information, describing the optimization points and potential adjustment suggestions of resource allocation, and providing basic data for further optimizing task execution.

[0041] Based on the resource rationality evaluation information, the task monitoring sub-module monitors the implementation of maintenance tasks, adjusts the operation process, matches the on-site situation and emergencies, and optimizes the task response and completion time, generating the resource allocation analysis result. Based on the resource rationality evaluation information, the execution of maintenance tasks is carefully monitored, including real-time tracking of task progress, analyzing the matching degree between the operation process and the actual situation, and making a quick response to emergencies. Through data analysis, the delays and inefficient links in task execution are identified, and accordingly, the operation process and resource allocation are adjusted, the task response and completion time are optimized, generating the resource allocation analysis result, and guiding the task scheduling and resource allocation strategy in the future time period.

[0042] The above is only the preferred embodiment of the present invention, and does not limit the present invention in other forms. Any person skilled in the art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An underground space infrastructure status monitoring and management system based on the Internet of Things, characterized in that, The system includes: The data acquisition and marking module collects multi-dimensional data of underground space infrastructure based on Internet of Things technology, preliminarily screens the data and marks timestamps, extracts key time points from the data, and generates a time series data set; The dynamic threshold adjustment module adjusts the moving average and exponentially weighted moving average of historical data based on the time series data set, performs data point matching, sets a dynamic threshold for groundwater level changes, and judges the abnormality of current data to generate an anomaly recognition index; The pattern analysis and learning module learns normal and abnormal patterns from the multi-dimensional data of underground space infrastructure using an LSTM model based on the anomaly recognition index, and analyzes the current data according to the patterns to generate a pattern recognition overview; The anomaly recognition and prediction module extracts key features from the pattern learning results, uses Seq2Seq to identify abnormal patterns, and predicts future potential problems and risks based on the identified features to generate a risk prediction summary; The task ranking and scheduling module ranks and schedules maintenance and inspection tasks using linear programming and integer programming based on the pattern recognition overview and risk prediction summary, and dynamically adjusts task priorities according to current and predicted situations to generate a scheduling optimization path; The maintenance task management module dynamically allocates maintenance tasks based on the scheduling optimization path according to the set priorities and real-time monitoring data, verifies the rationality of resource allocation in key underground areas, monitors the task execution process and adjusts the response process to generate a resource allocation analysis result.

2. The underground space infrastructure status monitoring and management system based on the Internet of Things according to claim 1, characterized in that: The time series data set includes data points, time tags, and key time intervals. The anomaly recognition index includes threshold level, anomaly score, and trend deviation information. The pattern recognition overview includes recognized patterns, anomaly degree, and normal behavior indicators. The risk prediction summary includes risk types, impact probabilities, and warning levels. The scheduling optimization path includes task numbers, priority levels, and estimated processing durations.

3. The state monitoring and management system for underground space infrastructure based on the Internet of Things according to claim 1, characterized in that The data acquisition and marking module includes: The data preliminary screening sub-module collects multi-dimensional data of underground space infrastructure based on Internet of Things technology, screens data with qualified signal strength, and eliminates weak or incorrect data points by evaluating signal quality to generate a valid data set; The timestamp marking sub-module assigns corresponding reception times to each piece of data based on the valid data set, and verifies the time information attached to each piece of data through time synchronization technology to generate a time index set; The key time point extraction sub-module analyzes data fluctuation points based on the time index set and marks the key time points of data changes, including equipment failure or abnormal operation times, to generate a time series data set.

4. The underground space infrastructure status monitoring and management system based on the Internet of Things according to claim 1, characterized in that The dynamic threshold adjustment module includes: The historical data analysis sub-module calculates data statistical indicators, including moving average and exponentially weighted moving average, based on the time series data set, smooths the fluctuations of time series data to generate a smooth time series; The threshold setting sub-module dynamically adjusts the threshold based on the smooth time series according to historical data and the current environment, sets a data threshold matching the current conditions, and generates threshold adjustment parameters; The anomaly judgment sub-module monitors the current data based on the threshold adjustment parameter, identifies abnormal or out-of-threshold data points, marks and records them, and generates an anomaly identification metric.

5. The underground space infrastructure status monitoring and management system based on the Internet of Things according to claim 1, characterized in that The pattern analysis and learning module includes: The pattern learning sub-module analyzes the multi-dimensional data of the underground space infrastructure based on the anomaly identification metric using the LSTM model, identifies the differences in behavior patterns in the time series, classifies the normal and abnormal behaviors of the data set, and generates a behavior pattern classification. The pattern analysis sub-module conducts a matching analysis of the data pattern based on the behavior pattern classification, compares the current data with the learned patterns, identifies the pattern deviations in the data, and generates pattern deviation identification information. The overview generation sub-module organizes the data pattern information based on the pattern deviation identification information, including the identification information of the data pattern and the behavior analysis results, and analyzes based on the current data to obtain a pattern recognition overview.

6. The underground space infrastructure status monitoring and management system based on the Internet of Things according to claim 1, characterized in that, The anomaly identification and prediction module includes: The feature extraction sub-module extracts key features from the pattern learning results, uses Seq2Seq to analyze the behavior features of abnormal and normal data, conducts key feature screening, identifies and extracts the key variables affecting pattern judgment, and generates a set of key variables. The pattern recognition sub-module conducts a sequence analysis of the data based on the set of key variables, identifies the abnormal behavior patterns in the data through behavior pattern matching, classifies and records the abnormal behaviors, and generates abnormal behavior identification information. The risk prediction sub-module conducts a trend analysis of the identified abnormal behavior patterns and evaluates the potential risks based on the abnormal behavior identification information, predicts the future potential risk development direction, and generates a risk prediction summary.

7. The underground space infrastructure status monitoring and management system based on the Internet of Things according to claim 6, characterized in that, According to the Seq2Seq, identify and extract the variables affecting pattern judgment from the abnormal and normal data behaviors, according to the formula: ； Calculate the feature score for each data point , where represents the feature weight, represents the corresponding feature value, represents the number of features.

8. The underground space infrastructure status monitoring and management system based on the Internet of Things according to claim 1, characterized in that, The task sorting and scheduling module includes: The task priority adjustment sub-module analyzes the risk levels and urgencies of multiple tasks based on the pattern recognition overview and the risk prediction summary, adjusts the priorities of maintenance and inspection tasks, updates the priorities of high-risk areas in real time, and generates a priority update list. The linear programming sub-module calculates the optimal resource allocation and time management process based on the priority update list through the linear programming algorithm, adjusts the task sequence to match the current available resources and time window, optimizes the task execution order, and generates an optimal task sequence. The integer programming sub-module conducts task scheduling based on the optimal task sequence through the integer programming algorithm, combines resource limitations and task divisibility, executes the optimal path of the task, and generates a scheduling optimization path.

9. The state monitoring and management system for underground space infrastructure based on the Internet of Things according to claim 8, characterized in that, According to the linear programming algorithm, calculate the optimal resource allocation and time management process, match the current available resources and time window, according to the formula: ； Minimize the computing resource cost, where represents the total cost, represents the unit price of the th resource item, represents the quantity of resources allocated to the th task item, represents the number of tasks.

10. The underground space infrastructure status monitoring and management system based on the Internet of Things according to claim 1, characterized in that, The maintenance task management module includes: The task assignment sub-module dynamically configures maintenance personnel and equipment to the critical operation areas based on the scheduling optimization path, combines real-time monitoring data and priorities, optimizes the maintenance efficiency and response speed, and generates a maintenance task configuration. The resource verification sub-module analyzes the resource usage and allocation efficiency in real time based on the maintenance task configuration, verifies the rationality of the resource configuration in the critical underground areas, and generates resource rationality evaluation information. Based on the resource rationality evaluation information, the task monitoring sub-module monitors the implementation of maintenance tasks, adjusts the operation process, matches the on-site situation and emergencies, optimizes the task response and completion time, and generates a resource allocation analysis result.

Images