Urban subway pipe network flood recognition system based on water level prediction and water seepage recognition

Abstract

The invention provides an urban subway pipe network flood recognition system based on water level prediction and water seepage recognition, and the system is characterized in that a sensor abnormality detection module carries out the prediction through an LSTM prediction model according to environment data, carries out the error recognition of a prediction result and real water level data, and obtains an abnormal water level detection point; the image anomaly detection module identifies a water seepage area from the image data by using contrast and brightness; the positioning module is used for positioning the water seepage position according to the abnormal water level detection point and the water seepage area; the flood risk chain construction module uses a PageRank algorithm to provide risk reference; the trend analysis module uses a seasonal decomposition algorithm to improve subway flood prediction accuracy; and the abnormity judgment module, the early warning module and the emergency disposal module are used for carrying out abnormity detection, early warning generation and emergency response on the data. In a word, the method can accurately identify the water seepage position in the subway, and early warning and response are carried out.

Description

Technical Field

[0001] The invention belongs to the field of urban subway pipe network flood analysis, and in particular relates to an urban subway pipe network flood identification system based on water level prediction and water seepage identification. Background Art

[0002] With the acceleration of urbanization, subways have become an important part of modern cities. However, there are many challenges in the construction and operation of subway pipelines, among which flooding is particularly prominent. Urban subway pipelines are often located deep underground and are vulnerable to waterlogging caused by groundwater leakage and heavy rainfall. The subway pipeline system contains multiple interconnected subsystems, such as drainage systems, communication systems, and power supply systems. It is in a dynamically changing environment and is affected by multiple factors such as groundwater levels, rainfall, and geological conditions. Changes in these factors may lead to varying degrees of risks in the subway pipeline system, which not only threatens the safety and stability of the subway structure, but also poses a serious threat to the normal operation of the subway.

[0003] Traditional flood monitoring and prevention methods rely on manual inspections and sensor monitoring, which cannot solve the systemic problems of subway pipe networks and cannot capture water seepage and water accumulation information in a timely and comprehensive manner. Moreover, subway pipe networks involve the collaborative work of multiple departments and units, and the management and maintenance tasks are arduous and complex. Under the existing disaster monitoring technology, it is difficult to accurately predict and prevent waterlogging accidents in advance, resulting in slow emergency response and increased accident risks. Especially in heavy rainfall or sudden floods, the identification and treatment of floods in subway pipe networks face huge challenges.

[0004] In conclusion, there is still much room for improvement in the existing flood detection and identification in urban subway systems. Summary of the invention

[0005] The present invention is made to solve the above-mentioned problems, and aims to provide a city subway pipe network flood identification system based on water level prediction and water seepage identification.

[0006] The present invention provides an urban subway pipe network flood identification system based on water level prediction and seepage identification, which has the following characteristics and includes: a data acquisition module, including multiple sensors and multiple cameras arranged at various locations in the subway, for real-time acquisition of environmental data and image data; a data sorting module, for preprocessing the environmental data and image data respectively, and obtaining preprocessed environmental data and preprocessed image data respectively; a data fusion module, for integrating the preprocessed environmental data and the preprocessed image data, and obtaining integrated data; a data verification module, for verifying the integrated data, and obtaining verification data; a sensor anomaly detection module, for identifying an abnormal water level detection point according to environmental data in the verification data; an image anomaly detection module, for identifying a seepage area according to image data in the verification data; and a positioning module, for locating the seepage position according to the abnormal water level detection point and the seepage area.

[0007] The urban subway pipe network flood identification system based on water level prediction and water seepage identification provided by the present invention may also have the following features: the sensor is a distributed optical fiber sensor, the environmental data includes water level data, temperature data and humidity data, and the camera is an ultra-high-definition camera.

[0008] The urban subway pipe network flood identification system based on water level prediction and seepage identification provided by the present invention may also have the following features: wherein, in the data sorting module, preprocessing includes noise cleaning, data structure conversion and missing data repair.

[0009] The urban subway pipe network flood identification system based on water level prediction and seepage identification provided by the present invention may also have the following features: wherein the data fusion module sequentially performs entity alignment, weighted fusion, spatial interpolation and multivariate linear regression calculations on the preprocessed environmental data and the preprocessed image data to obtain integrated data, the spatial interpolation is ArcGIS interpolation, and the multivariate linear regression calculations are performed through the spatiotemporal convolutional neural network.

[0010] The urban subway pipeline flood identification system based on water level prediction and seepage identification provided by the present invention may also have the following features: wherein, the data verification module verifies the integrated data through a dynamic time adjustment algorithm, and the data verification module distributes the data verification tasks to multiple nodes for parallel processing through a highly robust noise filtering algorithm and a dynamic weight allocation mechanism combined with a distributed operation structure.

[0011] The urban subway pipe network flood identification system based on water level prediction and seepage identification provided by the present invention may also have the following features: wherein the sensor anomaly detection module includes: a model prediction unit, including an LSTM prediction model, for inputting environmental data in the verification data into the LSTM prediction model to obtain water level prediction data; an abnormal point identification unit, for performing error calculation between the water level prediction data and the corresponding real water level data in the environmental data, and taking the detection point whose error solution result is greater than the error threshold as the abnormal water level detection point.

[0012] The urban subway pipe network flood identification system based on water level prediction and water seepage identification provided by the present invention may also have the following features: wherein the image anomaly detection module includes: a contrast calculation unit, which is used to calculate the average value of the absolute difference between the grayscale values ​​of the central pixel and the surrounding pixels as the contrast for the image data in the verification data; a water seepage identification unit, which stores a brightness threshold and a contrast threshold, and is used to take the area in the image data where the pixel brightness is less than the brightness threshold and the contrast is greater than the contrast threshold as the water seepage area.

[0013] The urban subway pipeline flood identification system based on water level prediction and seepage identification provided by the present invention may also have such features, including: a trend analysis module, which is used to decompose the verification data through a seasonal decomposition algorithm to obtain trend components, seasonal components and residual components, and obtain trend prediction analysis results based on the trend components and seasonal components, and obtain random fluctuations and abnormal point detection results based on the residual components.

[0014] In the urban subway pipeline flood identification system based on water level prediction and seepage identification provided by the present invention, it can also have such characteristics, including: a flood risk chain construction module and a deep learning analysis module, the deep learning analysis module includes a convolutional neural network, which is used to extract multiple features from the verification data as risk factors, and the flood risk chain construction module includes: a risk factor evaluation unit, which is used to calculate the PageRank value of each risk factor through the PageRank algorithm; a risk chain construction unit, which is used to identify key nodes and secondary nodes according to each PageRank value, and calculate the influence degree of each risk factor through the fuzzy comprehensive evaluation method, and generate a risk chain through the Markov chain method.

[0015] Functions and Effects of the Invention

[0016] According to the urban subway pipe network flood identification system based on water level prediction and water seepage identification involved in the present invention, on the one hand, through the sensor abnormality detection module, the LSTM prediction model is used to generate water level prediction data according to real-time data, and the water level prediction data and the real water level data are used to calculate the error, and the monitoring point with the error exceeding the threshold is used as the abnormal water level monitoring point; on the other hand, the image abnormality detection module uses contrast and brightness to identify the seepage area in the image, and then the positioning module combines the abnormal water level monitoring point and the seepage area to obtain the seepage position. Therefore, the urban subway pipe network flood identification system based on water level prediction and water seepage identification of the present invention can accurately identify the seepage position in the subway. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a block diagram of a system for identifying floods in urban subway pipe networks based on water level prediction and water seepage identification in an embodiment of the present invention;

[0018] Figure 2 is a schematic diagram of the flow of the PageRank algorithm in an embodiment of the present invention;

[0019] Figure 3 It is a schematic diagram of the process of constructing a risk chain in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the following embodiments and the accompanying drawings specifically illustrate the urban subway pipeline flood identification system based on water level prediction and water seepage identification of the present invention.

[0021] This embodiment provides an urban subway pipe network flood identification system based on water level prediction and water seepage identification, which is used to identify the water seepage location in the subway.

[0022] Figure 1 It is a block diagram of a city subway pipe network flood identification system based on water level prediction and water seepage identification in an embodiment of the present invention.

[0023] like Figure 1 As shown, the urban subway pipeline flood identification system 100 based on water level prediction and seepage identification includes a data acquisition module 11, a data sorting module 12, a data fusion module 13, a data verification module 14, a sensor anomaly detection module 15, an image anomaly detection module 16, a positioning module 17, a trend analysis module 18, a deep learning analysis module 19, a flood risk chain construction module 20, an anomaly judgment module 21, an early warning module 22, an emergency response module 23 and a control module 24 that controls the operation of the above modules.

[0024] The data acquisition module 11 includes a plurality of sensors and a plurality of cameras arranged at various locations in the subway, and is used to collect environmental data and image data in real time.

[0025] Among them, the sensor is a distributed optical fiber sensor, and the camera is an ultra-high-definition camera. Environmental data includes water level data, temperature data, and humidity data. In this embodiment, the distributed optical fiber sensors are distributed in various locations of the subway tunnel, and have good anti-interference capabilities, and can achieve long-distance coverage and high-speed response. Multiple ultra-high-definition cameras are respectively arranged in various pipe networks of the subway, and have high-resolution camera capabilities, and also support local storage function, cloud storage function, and low-power sleep / wake-up function.

[0026] The data sorting module 12 is used to preprocess the environment data and the image data to obtain preprocessed environment data and preprocessed image data respectively.

[0027] In the data sorting module 12, preprocessing includes noise cleaning, data structure conversion and missing data repair. In this embodiment, the data sorting module 12 is used to solve the noise and data missing problems that are unique to the subway environment.

[0028] The data fusion module 13 is used to integrate the pre-processed environment data and the pre-processed image data to obtain integrated data.

[0029] The data fusion module 13 sequentially performs entity alignment, weighted fusion, spatial interpolation and multiple linear regression calculation on the pre-processed environment data and the pre-processed image data to obtain integrated data. The spatial interpolation is ArcGIS interpolation. The multiple linear regression calculation is performed through the spatiotemporal convolutional neural network.

[0030] In this embodiment, the data fusion module 13 fuses multi-source data in time, space and multiple dimensions. By combining data from multiple time periods and multiple sources, a clearer and more comprehensive understanding of the overall system, namely, integrated data, can be formed.

[0031] The data verification module 14 is used to verify the integrated data to obtain verification data.

[0032] The data verification module 14 verifies the integrated data through a dynamic time adjustment algorithm. The data verification module 14 distributes the data verification task to multiple nodes for parallel processing through a highly robust noise filtering algorithm and a dynamic weight distribution mechanism combined with a distributed operation structure.

[0033] The sensor abnormality detection module 15 is used to identify abnormal water level detection points according to the environmental data in the verification data.

[0034] The sensor anomaly detection module 15 includes a model prediction unit 151 and an anomaly identification unit 152 .

[0035] The model prediction unit 151 includes an LSTM prediction model, which is used to input the environmental data in the verification data into the LSTM prediction model to obtain water level prediction data.

[0036] In this embodiment, the LSTM prediction model can process time series data, learn the long-term patterns and trends of water level data, thereby identifying sudden abnormal rises in water levels, and then automatically identify abnormal patterns and emergencies in the data.

[0037] The abnormal point identification unit 152 is used to calculate the error between the water level prediction data and the corresponding real water level data in the environmental data, and take the detection point whose error solution result is greater than the error threshold as the abnormal water level detection point.

[0038] The image anomaly detection module 16 is used to identify the water seepage area according to the image data in the verification data.

[0039] The image anomaly detection module includes a contrast calculation unit 161 and a water seepage identification unit 162 .

[0040] The contrast calculation unit 161 is used to calculate the average value of the absolute difference between the grayscale values ​​of the central pixel and the surrounding pixels as the contrast for the image data in the verification data.

[0041] The water seepage identification unit 162 stores a brightness threshold and a contrast threshold, and is used to regard a region in the image data where the pixel brightness is less than the brightness threshold and the contrast is greater than the contrast threshold as a water seepage region.

[0042] The positioning module 17 is used to locate the water seepage location according to the abnormal water level detection point and the water seepage area.

[0043] The trend analysis module 18 is used to decompose the verification data through the seasonal decomposition algorithm to obtain trend components, seasonal components and residual components, and obtain trend prediction analysis results based on the trend components and seasonal components, and obtain random fluctuations and outlier detection results based on the residual components. In this embodiment, the trend component can adapt to system characteristics that change over time and provide continuous monitoring and analysis of dynamic problems. The seasonal component can identify periodic changes and has better adaptability and prediction capabilities. The residual component can capture irregular fluctuations in the system that change over time, so that the system can monitor and respond to abnormal situations in real time.

[0044] In this embodiment, the water level data, water seepage data, and environmental data in the subway network are processed by the trend analysis module 18 to analyze the seasonal variation characteristics of these data, thereby obtaining more accurate flood prediction results under different seasons and climate conditions.

[0045] The deep learning analysis module 19 includes a convolutional neural network for extracting multiple features from the verification data as risk factors.

[0046] The convolutional layer of the convolutional neural network in this embodiment can automatically extract representative features from different data sources, such as temperature, humidity, seasonal changes, geographic location, etc. Using deep learning technology, the convolutional neural network conducts comprehensive analysis of various environmental data and identifies complex risk patterns. Through the systematic analysis of multi-source data, a solid foundation is provided for the construction of the flood risk chain, revealing potential systemic risk factors and their interrelationships.

[0047] The flood risk chain construction module 20 includes a risk factor assessment unit 201 and a risk chain construction unit 202 .

[0048] The risk factor evaluation unit 201 is used to calculate the PageRank value of each risk factor through the PageRank algorithm.

[0049] In this embodiment, each risk factor is used as a node to construct a network, and the relationship between risk factors is the edge in the network. The PageRank algorithm can be used to evaluate the relative importance of these nodes, help identify the most critical factors in the risk chain, and provide effective risk assessment results. The algorithm can reveal the importance of each risk node in the entire subway pipe network system, identify key disaster-causing factors and their chain reaction paths, and thus take targeted preventive measures. The PageRank algorithm expresses and evaluates risk factors in a networked manner, and the results of the risk factor importance assessment provide important decision-making basis for a series of subsequent optimization management, resource allocation, emergency response, and system improvement. It can focus resource management on the areas with the greatest impact and achieve greater decision-making benefits under limited resource conditions.

[0050] Figure 2 It is a flow chart of the PageRank algorithm in an embodiment of the present invention.

[0051] like Figure 2 As shown, the process of calculating the PageRank value of each risk factor through the PageRank algorithm includes the following steps:

[0052] Step S1, constructing a risk factor set F and an association relationship set E according to various risk factors and their mutual associations.

[0053] Step S2, constructing a directed graph G according to the risk factor set F and the association relationship set E.

[0054] Step S3, initializing each node of the directed graph G. In this embodiment, the initial PageRank value of each node is r=1 / n, where n is the total number of nodes.

[0055] Step S4, calculate and update the PageRank value of each node according to the PageRank formula, and iterate repeatedly until the PageRank values ​​of all nodes converge.

[0056] In this embodiment, the higher the PageRank value of the risk factor is, the more important it is in the risk analysis, indicating that its impact on the flood is greater.

[0057] The risk chain construction unit 202 is used to identify key nodes and secondary nodes according to each PageRank value, calculate the influence degree of each risk factor through the fuzzy comprehensive evaluation method, and generate a risk chain through the Markov chain method.

[0058] In this embodiment, the risk chain construction unit 202 stores a plurality of preset risk evaluation levels, which are high risk, medium-high risk, medium risk, medium-low risk, and low risk. The risk chain construction unit 202 comprehensively considers each risk factor through the fuzzy comprehensive evaluation method, and then uses the fuzzy center of gravity method to comprehensively consider the membership and weight to quantify the influence of each risk factor.

[0059] The Markov chain method is to construct the state transition probability matrix of the Markov chain based on historical data and expert evaluation to simulate the transfer of risks between different states. After iteration, the state distribution converges to a stable state, and the probability of the system under various risk factors can be obtained, thereby quantifying the risk level of the overall subway network system under various risk factors, that is, the risk assessment level.

[0060] Figure 3 It is a schematic diagram of the process of constructing a risk chain in an embodiment of the present invention.

[0061] like Figure 3 As shown, the process of risk chain building unit 202 building a risk chain includes the following steps:

[0062] Step T1, construct multiple risk assessment levels. In this embodiment, the risk assessment levels are from high to low: high risk, medium-high risk, medium risk, medium-low risk, and low risk.

[0063] Step T2, comprehensively consider various risk factors through fuzzy comprehensive evaluation method and construct the fuzzy relationship matrix R.

[0064] Step T3, calculate the weight vector A of each risk factor through the hierarchical analysis method.

[0065] Step T4, multiply the weight vector A by the fuzzy relationship matrix R to obtain a comprehensive evaluation result vector B.

[0066] Step T5, using the fuzzy centroid method, comprehensively consider the membership and weight to obtain the impact of each risk factor.

[0067] In this embodiment, the comprehensive evaluation result vector B is operated with the matrix formed by the risk evaluation level to obtain the fuzzy comprehensive value G, which is calculated as follows:

[0068]

[0069] Where b i is the value corresponding to risk factor i in the comprehensive evaluation result vector B, v i is the value corresponding to the risk assessment level of risk factor i, for example, high risk, medium-high risk, medium risk, medium-low risk, and low risk are set to 50, 40, 30, 20, and 10 respectively.

[0070] In this embodiment, a sensitivity analysis method is used to change the weight vector A corresponding to the risk factor, thereby calculating the corresponding fuzzy comprehensive value G, and the influence degree of each risk factor is obtained by analyzing the numerical change of the fuzzy comprehensive value G.

[0071] Step T6, determine the risk state and state space. In this embodiment, the centralized risk state and state space include all possible risk states and transition paths.

[0072] Step T7, constructing a state transition probability matrix P based on historical data and transition probabilities.

[0073] Step T8, iterate the state transition probability matrix P until the state distribution converges to a stable state, and then generate a risk chain based on the state transition matrix and the final state distribution.

[0074] The abnormality judgment module 21 is used to generate an abnormality detection result according to the risk chain, random fluctuation and abnormal point detection results, water seepage location, abnormal water level detection point, environmental data and image data. The abnormality detection result is no abnormality or abnormality exists.

[0075] The warning module 22 is used to generate warning information and send the warning information to relevant personnel when the abnormality detection result shows that there is an abnormality.

[0076] The emergency handling module 23 is used to generate corresponding emergency handling steps and resource allocation plans when the abnormality detection result shows that an abnormality exists.

[0077] The control module 24 stores a control program for controlling the operation of each module.

[0078] Functions and Effects of the Embodiments

[0079] According to the urban subway pipe network flood identification system based on water level prediction and seepage identification involved in this embodiment, on the one hand, through the sensor anomaly detection module, the LSTM prediction model is used to generate water level prediction data based on real-time data, and the water level prediction data is compared with the real water level data for error calculation, and the monitoring point with an error exceeding the threshold is used as an abnormal water level monitoring point; on the other hand, the image anomaly detection module uses contrast and brightness to identify the seepage area in the image, and then the positioning module combines the abnormal water level monitoring point and the seepage area to obtain the seepage position. In short, this method can accurately identify the seepage position in the subway.

[0080] Furthermore, the seasonal variation characteristics of water level data, seepage data and environmental data are obtained through the trend analysis module using the seasonal decomposition algorithm to improve the accuracy of subway flood prediction.

[0081] Furthermore, the PageRank algorithm is used to calculate the PageRank value of risk factors through the flood risk chain construction module, and then the Markov chain method is used to generate the risk chain to provide risk reference.

[0082] Furthermore, the real-time data and processed data are further processed by the abnormal judgment module to obtain abnormal detection results. And through the early warning module and emergency response module, early warning information notification and early warning plan generation are carried out for abnormal results, so as to realize timely judgment and response to subway floods.

[0083] Those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A flood identification system for urban subway pipe networks based on water level prediction and water seepage identification, used to identify water seepage locations in subways, characterized in that: include: The data collection module includes multiple sensors and multiple cameras installed in various places of the subway, which are used to collect environmental data and image data in real time; A data sorting module, used for preprocessing the environment data and the image data to obtain preprocessed environment data and preprocessed image data respectively; A data fusion module, used for integrating the preprocessed environment data and the preprocessed image data to obtain integrated data; A data verification module, used to verify the integrated data to obtain verification data; A sensor anomaly detection module, used to identify an abnormal water level detection point according to the environmental data in the verification data; An image anomaly detection module, used to identify a water seepage area based on the image data in the verification data; A positioning module is used to locate the water seepage location according to the abnormal water level detection point and the water seepage area.

2. The urban subway pipe network flood identification system based on water level prediction and water seepage identification according to claim 1 is characterized by: in, The sensor is a distributed optical fiber sensor, The environmental data includes water level data, temperature data and humidity data, The camera is an ultra-high-definition camera.

3. The urban subway pipe network flood identification system based on water level prediction and water seepage identification according to claim 1 is characterized by: in, In the data sorting module, the preprocessing includes noise cleaning, data structure conversion and missing data repair.

4. The urban subway pipe network flood identification system based on water level prediction and water seepage identification according to claim 1 is characterized by: in, The data fusion module sequentially performs entity alignment, weighted fusion, spatial interpolation and multivariate linear regression calculation on the pre-processed environment data and the pre-processed image data to obtain the integrated data. The spatial interpolation is ArcGIS interpolation. The multivariate linear regression calculation is performed by a spatiotemporal convolutional neural network.

5. The urban subway pipe network flood identification system based on water level prediction and water seepage identification according to claim 1 is characterized by: in, The data verification module verifies the integrated data through a dynamic time adjustment algorithm. The data verification module distributes the data verification tasks to multiple nodes for parallel processing through a highly robust noise filtering algorithm and a dynamic weight allocation mechanism combined with a distributed operation structure.

6. The urban subway pipe network flood identification system based on water level prediction and water seepage identification according to claim 1, Features: in, The sensor anomaly detection module comprises: A model prediction unit, comprising an LSTM prediction model, for inputting environmental data in the verification data into the LSTM prediction model to obtain water level prediction data; The abnormal point identification unit is used to calculate the error between the water level prediction data and the corresponding real water level data in the environmental data, and take the detection point whose error solution result is greater than the error threshold as the abnormal water level detection point.

7. The urban subway pipe network flood identification system based on water level prediction and water seepage identification according to claim 1, Features: in, The image anomaly detection module comprises: A contrast calculation unit, used to calculate the average value of the absolute difference between the grayscale values ​​of the central pixel and the surrounding pixels as the contrast for the image data in the verification data; The water seepage identification unit stores a brightness threshold and a contrast threshold, and is used to take the area in the image data where the pixel brightness is less than the brightness threshold and the contrast is greater than the contrast threshold as the water seepage area.

8. The urban subway pipe network flood identification system based on water level prediction and water seepage identification according to claim 1 is characterized in that: Also includes: The trend analysis module is used to decompose the verification data through a seasonal decomposition algorithm to obtain a trend component, a seasonal component and a residual component, and obtain a trend prediction analysis result based on the trend component and the seasonal component, and obtain a random fluctuation and anomaly detection result based on the residual component.

9. The urban subway pipe network flood identification system based on water level prediction and water seepage identification according to claim 1 is characterized in that: Also includes: Flood risk chain building module and deep learning analysis module, The deep learning analysis module includes a convolutional neural network, which is used to extract multiple features from the verification data as risk factors, The flood risk chain building modules include: A risk factor evaluation unit, used to calculate the PageRank value of each risk factor through the PageRank algorithm; The risk chain construction unit is used to identify key nodes and secondary nodes according to each of the PageRank values, calculate the influence degree of each of the risk factors through a fuzzy comprehensive evaluation method, and generate a risk chain through a Markov chain method.

Images