A real-time analysis method for urban sewer network clogging based on street monitoring

Abstract

The application provides a kind of city drainage pipe network clogging real-time analysis method based on street monitoring, to solve the problem that city rainwater pipe network siltation is difficult to be found in time, siltation position and degree are difficult to determine, professional pipe network monitoring equipment is expensive, and real-time analysis and diagnosis are difficult, the present application combines existing city street monitoring and convolutional neural network (CNN) real-time calculation flood situation, using city rain flood model, monte carlo method, back propagation neural network (BP) coupling model for pipe network siltation and position real-time judgment analysis, in addition, the city drainage pipe network clogging real-time analysis method based on street monitoring provided by the application does not need professional equipment, does not need to add hardware equipment, and can realize real-time calculation of city rainwater pipe network siltation position and degree prediction.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of urban rain flood simulation, and particularly relates to a real-time analysis method for urban drainage pipe network blockage based on street monitoring. BACKGROUND

[0002] The urban rainwater pipe network system is prone to blockage during operation, which leads to a decrease in the pipe network drainage capacity, and further causes problems such as poor drainage during flood season, road surface waterlogging and even urban waterlogging.

[0003] The existing rainwater pipe network blockage detection methods mainly include closed-circuit television (CCTV) detection, quick video detection (QV) detection and sonar detection, but these methods require professional equipment and operators, and the period is long and the cost is high when comprehensively detecting long-distance or large-area pipe networks. In addition, these methods rely on periodic inspection and cannot monitor the dynamic changes of the pipe in real time, and the response to sudden blockage is not flexible enough. Rainwater pipe blockage is a dynamic process, especially during heavy rainfall, branches, leaves, household garbage and sediment blocks may cause rapid blockage of the pipe network, and even pipe structure instability and collapse, resulting in complete loss of drainage function. The traditional detection method is difficult to analyze and diagnose the pipe network blockage position and degree in real time. SUMMARY

[0004] The purpose of the present application is to provide a real-time analysis method for urban drainage pipe network blockage based on street monitoring, in order to provide a new idea for solving real-time prediction and analysis of rainwater pipe network blockage.

[0005] To this end, the above-mentioned purpose of the present application is realized by the following technical scheme:

[0006] A real-time analysis method for urban drainage pipe network blockage based on street monitoring, comprising the following steps:

[0007] S1, arranging basic data and establishing a pipe network-ground double-layer urban rain flood model;

[0008] S2, using the accumulated water condition of the measured rainfall to complete the calibration of the pipe network-ground double-layer urban rain flood model;

[0009] S3, setting pipe network blockage and rainfall working condition parameters, and using the Monte Carlo method to generate different blockage positions and rainfall data sets for the pipe network-ground double-layer urban rain flood model;

[0010] S4, inputting the data set into the pipe network-ground double-layer urban rain flood model to simulate the ground waterlogging condition when the pipe network and the rainwater inlet are blocked at different positions;

[0011] S5, training a BP neural network using the pipe network blockage condition and the ground waterlogging data set;

[0012] S6, using the monitoring video of the street waterlogging during heavy rain and the elevation and width of the reference point, training the CNN neural network;

[0013] S7, inputting the street monitoring video into the CNN neural network to obtain the flooding situation of different ground positions;

[0014] S8, inputting the flooding situation of the ground into the BP neural network to obtain the siltation situation of different nodes.

[0015] While the above technical solutions are adopted, the application can also adopt or combine the following technical solutions:

[0016] As a preferred technical solution of the application: in step S1, the basic data includes rainwater pipe network data, ground elevation data, land use data, remote sensing image data and design rainfall data.

[0017] As a preferred technical solution of the application: in step S1, the pipe network-ground double-layer city rain flood model includes an underground rainwater pipe network module, a ground street and waterlogging point module, a rainwater pump station module and a rainwater storage tank module.

[0018] As a preferred technical solution of the application: in step S2, collect and measure the rainwater pipe network siltation and blockage status data, and reflect it in the city rain flood model; collect and observe the corresponding measured rainfall process and ground flooding situation, and rate the parameters of the city rain flood model according to the measured data, so that the city rain flood model can restore the city ground waterlogging situation during heavy rain with high precision.

[0019] As a preferred technical solution of the application: in step S3, set the pipe network and rainwater inlet siltation working condition, set the flow section blockage parameter value to 0%, 30%, 50%, 70% and 90% four working conditions, and set the probability of each parameter value to 60%, 20%, 10%, 5% and 5% respectively; at the same time, generate 2-hour rainfall processes of six Chicago rain types P=1a, 3a, 5a, 10a, 20a and 50a; use the Monte Carlo method to randomly generate a large number of parameter sample data sets according to the above data distribution law.

[0020] As a preferred technical solution of the application: in step S4, according to the different siltation positions and rainfall data sets generated in step S3, set different siltation conditions of each pipe network and rainwater inlet of the city rain flood model, and perform simulation calculation to generate ground waterlogging data corresponding to the pipe network siltation; during calculation, multi-core parallel technology is adopted, the generated ground waterlogging data is combined with the pipe network and rainwater inlet siltation data of the previous step to form a combined data set of siltation and corresponding waterlogging situation.

[0021] As a preferred technical solution of the present application: in step S5, the n independent BP neural networks are trained using the combined data set generated in step S4, and n corresponds to the total number of pipes that are focused on in the urban flood model; the input data of the neural network is: the water depth of the surrounding m water accumulation points, and the rainfall intensity; the output data of the neural network is: the blockage coefficient of the pipe.

[0022] As a preferred technical solution of the present application: in step S6, the input data of the CNN neural network is: a single frame image of the monitoring video when the storm street is flooded, and the image is processed for denoising, enhancement and normalization before inputting the model to improve the quality; the reference point with known height in the artificially labeled image; the output data of the CNN neural network is: the water depth information of the single frame image of the monitoring video.

[0023] As a preferred technical solution of the present application: in step S7, the street monitoring information of the key concerned area is collected and analyzed in real time; in order to improve the working efficiency of the model and reduce the resource consumption of the system, the key frames of the collected monitoring video are analyzed, specifically, one frame is taken every 1-10 minutes, and the corresponding time interval should be associated with the rainfall information, and the larger the rainfall, the shorter the interval.

[0024] As a preferred technical solution of the present application: in step S8, the water depth data output by the CNN model is analyzed, the water depth data is input into the BP neural network corresponding to the pipe or the rainwater inlet, and the blockage of the corresponding node is calculated.

[0025] The present application provides a kind of based on street monitoring urban drainage pipe network real-time analysis method of silt, to solve the problem that urban rainwater pipe network siltation blockage is difficult to be discovered in time, siltation blockage position and degree are difficult to determine, professional pipe network monitoring equipment is expensive, difficult to real-time analysis and diagnosis, the present application combines existing urban street monitoring and convolutional neural network (CNN) real-time calculation flood situation, utilizes urban rain flood model, Monte Carlo method, back propagation neural network (BP) coupling model to carry out pipe network siltation and position real-time judgment analysis, the method proposed in the present application does not need professional equipment, does not need to add hardware equipment, can achieve the real-time calculation of urban rainwater pipe network siltation position and degree prediction.

[0026] The basic principle of the present application is that rainwater pipe blockage will increase the surface flooding range and depth, and generally, the surface water accumulation degree of the area near the corresponding blocked position will increase with the increase of the pipe blockage degree. That is, under different rainfall scenarios, there is a nonlinear function relationship between the surface water accumulation and the pipe blockage degree of the corresponding position. The present application constructs a large sample parameter data set of pipe network blockage under different conditions by using the urban rain flood model and the Monte Carlo method, and uses the BP neural network to learn and extract the correlation between the pipe blockage degree and the surface water accumulation in the data set, and at the same time, uses the CNN neural network to extract the flooded water depth information in the street monitoring during the rainstorm. During the system operation, the extracted surface flooding information is input into the trained BP neural network, and the blockage condition of the pipe section at the corresponding position can be obtained.

[0027] Compared with the prior art, the present application has the following beneficial effects:

[0028] (1) Dynamic real-time analysis of rainwater pipe network blockage position and degree.

[0029] Traditional rainwater pipe network blockage detection relies on periodic investigation, which requires professional personnel and equipment, is expensive and time-consuming. However, pipe network blockage is prone to occur during the flood season, and the blockage condition is sudden and hidden, but it has a huge impact on the drainage capacity of the pipe network, causing or exacerbating urban waterlogging during heavy rainfall. The traditional method is difficult to realize real-time perception and analysis of the blockage position and condition.

[0030] The present application can realize real-time analysis of the pipe network blockage point position and degree by using neural network and urban rain flood model based on the information of water accumulation point position, rainfall intensity, etc.

[0031] (2) No need to add hardware equipment, low cost.

[0032] Traditional methods rely on robots equipped with image sensors or sonar and other professional equipment to detect rainwater pipe network, or lay water level and flow sensors to monitor the drainage pipe network. The above solutions require the use of professional equipment and have a certain risk of damage, and the cost is relatively high. The present application uses existing urban street monitoring and does not need to purchase additional hardware equipment, which is low in cost. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 The operation flowchart of the urban drainage pipe network blockage real-time analysis method based on street monitoring provided by the present application is shown.

[0034] Figure 2 The step diagram of the urban drainage pipe network blockage real-time analysis method based on street monitoring provided by the present application is shown. DETAILED DESCRIPTION

[0035] The application will be further described in detail with reference to the accompanying drawings and specific embodiments.

[0036] The application provides a real-time analysis method for urban drainage network clogging based on street monitoring, comprising the following steps:

[0037] S1: Organize basic data and establish a "network-ground" double-layer urban rain flood model.

[0038] The basic data should include rainwater network data, ground elevation data, land use data, remote sensing image data and design rainfall data. The double-layer urban rain flood model established includes necessary modules such as underground rainwater network, ground street and prone to flooding points, rainwater pumping stations and rainwater storage tanks.

[0039] S2: Complete model calibration using the waterlogging situation of measured rainfall.

[0040] Collect and measure the current data of rainwater network clogging and reflect it in the urban rain flood model. Collect and observe the corresponding measured rainfall process and ground flooding situation, and calibrate the parameters of the urban rain flood model according to the measured data, so that the urban rain flood model can restore the urban ground waterlogging situation in heavy rain with high precision.

[0041] S3: Set the network clogging and rainfall working condition parameters, and use the Monte Carlo method to generate different clogging positions and rainfall data sets for the urban rain flood model.

[0042] Set the network and rainwater inlet clogging working conditions, such as setting the clogging parameter value of the flow section to 0%, 30%, 50%, 70% and 90% of four working conditions, and setting the probability of each parameter value to 60%, 20%, 10%, 5% and 5% respectively. At the same time, generate 2-hour rainfall processes of six Chicago rain types P=1a, 3a, 5a, 10a, 20a and 50a. Use the Monte Carlo method to randomly generate a large number of parameter sample data sets, such as 300 million parameter data sets.

[0043] S4: Input the data set into the urban rain flood model to simulate the ground waterlogging situation when the network and rainwater inlet are clogged at different positions.

[0044] According to the data set generated in the above step, set different clogging conditions of each network and rainwater inlet of the urban rain flood model, and perform simulation calculation to generate 300 million ground waterlogging data corresponding to network clogging. Multi-core parallel technology is used in the calculation. The generated ground waterlogging data is combined with the network and rainwater inlet clogging data of the previous step to form a combined data set of 300 million clogging conditions and corresponding waterlogging conditions.

[0045] S5: Train the BP neural network using the network clogging condition and ground waterlogging data set.

[0046] Using the combined data set generated in the previous step, train n an independent BP neural network, n The corresponding number is the total number of pipes that are the focus of the urban flood model.

[0047] The input data of the neural network is: the water depth of the surrounding m accumulation point, rainfall intensity; the output data of the neural network is: the blockage coefficient of the pipe.

[0048] S6: Use the monitoring video of the street waterlogging during the rainstorm and the elevation and width of the reference point to train the CNN neural network, so that the CNN neural network can output the street flooding depth with high precision.

[0049] The input data of the CNN neural network is: 1) the single frame image of the monitoring video when the street accumulates water during the rainstorm, and the image is denoised, enhanced, and normalized before inputting the model to improve the quality; 2) the reference points (such as curb stones, road signs, vehicle tires, etc.) with known height marked by artificial.

[0050] The output data of the CNN neural network is: the water depth information of the single frame image of the monitoring video.

[0051] S7: When the system starts running during the rainstorm, input the street monitoring video into the CNN neural network to obtain the flooding situation at different ground positions.

[0052] When the system is running, real-time street monitoring information of the key area is collected and analyzed. In order to improve the working efficiency of the model and reduce the resource consumption of the system, the key frames of the collected monitoring video are analyzed, specifically, one frame is taken every 1-10 minutes, and the corresponding time interval should be related to the rainfall information, the larger the rainfall, the shorter the interval.

[0053] After taking the interval frames of the monitoring video, the image is preprocessed, and the processed image is input into the CNN neural network to output the flooding water depth.

[0054] S8: When the system starts running during the rainstorm, input the flooding situation of the ground into the BP neural network to obtain the siltation situation of different nodes.

[0055] Analyze the water depth data output by the CNN model, input the water depth data into the BP neural network of the corresponding pipe or rainwater inlet, and calculate the blockage of the corresponding node.

[0056] The above specific embodiments are used to explain and illustrate the present application, and are only preferred embodiments of the present application, but not limit the present application, any modification, equivalent replacement, improvement, etc. of the present application within the spirit and protection scope of the claims of the present application, fall into the protection scope of the present application.

Claims

1. A real-time analysis method for urban sewer network clogging based on street monitoring, characterized in that, The method comprises the following steps: S1, collate basic data, and establish a pipe network-ground surface double-layer urban rain flood model; S2, use the accumulated water condition of the measured rainfall to complete the calibration of the pipe network-ground surface double-layer urban rain flood model; S3, set pipe network clogging and rainfall working condition parameters, and use the Monte Carlo method to generate different clogging positions and rainfall data sets for the pipe network-ground surface double-layer urban rain flood model; S4, input the data sets into the pipe network-ground surface double-layer urban rain flood model, and simulate the ground surface accumulated water condition when the pipe network and rainwater inlet at different positions are clogged; S5, use the pipe network clogging condition and the ground surface accumulated water data set to train a BP neural network; S6, use the monitoring video of the street accumulated water during the rainstorm and the datum point elevation and width to train a CNN neural network; S7, input the street monitoring video into the CNN neural network to obtain the submergence condition at different ground surface positions; S8, input the ground surface submergence condition into the BP neural network to obtain the clogging condition of different nodes. In step S5, the combined data set generated in step S4 is used to train n independent BP neural networks, and n corresponds to the total number of pipes that are focused on in the urban flood model; the input data of the neural network are: the water depth of the surrounding m accumulated water points and the rainfall intensity; and the output data of the neural network are: the clogging coefficient of the pipe.

2. The method of claim 1, wherein, In step S1, the basic data include rainwater pipe network data, ground surface elevation data, land use data, remote sensing image data and design rainfall data.

3. The method of claim 1, wherein, In step S1, the pipe network-ground surface double-layer urban rain flood model comprises an underground rainwater pipe network module, a ground surface street and prone-to-flooding point module, a rainwater pumping station module and a rainwater storage tank module.

4. The method of claim 1, wherein, In step S2, the rainwater pipe network clogging and blockage status data are collected and measured, and are reflected in the urban rain flood model; the corresponding measured rainfall process and ground surface submergence condition are collected and observed, and the parameters of the urban rain flood model are calibrated according to the measured data, so that the urban rain flood model can restore the urban ground surface accumulated water condition during the rainstorm with high precision.

5. The method of claim 1, wherein, In step S3, the pipe network and rainwater inlet clogging working condition is set, the clogging parameter value of the flow section is set to four working conditions of 0%, 30%, 50%, 70% and 90%, the probabilities of the parameter values are set to 60%, 20%, 10%, 5% and 5% respectively; meanwhile, six kinds of 2-hour rainfall processes of Chicago rain type P=1a, 3a, 5a, 10a, 20a and 50a are generated; a large number of parameter sample data sets are randomly generated according to the above data distribution law by using the Monte Carlo method.

6. The method of claim 1, wherein, In step S4, different clogging positions and rainfall data sets generated in step S3 are set, different clogging conditions of the pipe network and rainwater inlet of the urban rain flood model are simulated, and the ground surface accumulated water data corresponding to the pipe network clogging are generated; the multi-core parallel technology is adopted during the calculation, the ground surface accumulated water data generated are combined with the pipe network and rainwater inlet clogging data of the previous step to form a combined data set of the clogging condition and the corresponding accumulated water condition.

7. The method of claim 1, wherein, In step S6, the input data of the CNN neural network is a single frame image of the monitoring video when the heavy rain street is flooded, the image before inputting the model is processed by denoising, enhancing and normalizing to improve the quality; the reference point with known height in the artificially labeled image; the output data of the CNN neural network is the water depth information of the single frame image of the monitoring video.

8. The method of claim 1, wherein, In step S7, the street monitoring information of the area of focus is collected and analyzed in real time; the key frames of the collected monitoring video are analyzed, and one frame is intercepted every 1-10 minutes, and the corresponding time interval should be associated with the rainfall information, and the larger the rainfall, the shorter the interval.

9. The method of claim 1, wherein, In step S8, the water depth data output by the CNN model is analyzed, the water depth data is input into the BP neural network corresponding to the pipeline or the rainwater inlet, and the blockage of the corresponding node is calculated.

Images