Intelligent underground drainage pipe network management method and system

Abstract

The application discloses a kind of intelligent underground drainage pipe network management method and system, it is related to drainage pipe network management field, including obtaining the real-time liquid level and real-time flow of target pipe section upstream monitoring point;Monitoring data is divided into rising water period data and recession period data;Two-way resistance characteristic analysis is carried out to liquid level-flow model, resistance release characteristic and resistance recovery characteristic are identified, resistance release characteristic is in rising water period or rising recession junction liquid level reaches first threshold value when flow surge and liquid level rise amplitude stagnates or drops suddenly, resistance recovery characteristic is in recession period liquid level rebounds when flow reduces to second threshold value;When two kinds of characteristics are detected simultaneously, it is preliminarily judged that there is soft intermittent blockage anomaly in target pipe section;Flexible dredging management work order is generated based on the determination result, and asset state mark in pipe network operation and maintenance database is updated, the application solves the problem that soft intermittent blockage anomaly is difficult to identify and operate and maintain timely and accurately.

Description

Technical Field

[0001] This invention relates to the field of drainage network management, and more specifically, to an intelligent underground drainage network management method and system. Background Technology

[0002] With the deepening development of smart cities, the refined management of urban underground drainage pipe networks has become a core element in improving urban flood control capabilities and ensuring water environment safety. Modern pipe network management systems typically rely on IoT sensors deployed at key nodes to establish a digital monitoring platform by acquiring hydraulic parameters such as water level, flow rate, and flow velocity in real time. These systems aim to achieve real-time perception of pipe network operation status, automatic early warning of abnormal events, and scientific scheduling of operation and maintenance resources through the processing and analysis of massive amounts of monitoring data.

[0003] In existing technologies, the assessment and maintenance decisions for pipeline network operation status often rely on hydraulic model simulation or single-dimensional monitoring data threshold analysis. For example, Chinese patent application CN120806906A (A method for guiding precise maintenance and operation of urban pipeline networks using SWMM models) discloses a technology that simulates the pipeline network operation status by constructing an SWMM hydraulic model, comparing the simulation results with measured data to assess the degree of siltation, and thus formulating a maintenance plan. Another example is Chinese patent application CN114611728A (A method and system for monitoring blockages in sewage pipeline networks), which proposes a method for identifying blockages based on the changing trends of liquid level and flow data at monitoring nodes, primarily focusing on identifying continuous abnormal increases in water level or decreases in flow.

[0004] However, existing technologies have significant limitations in handling specific types of intermittent anomalies. In actual pipeline network operation, there exists a type of intermittent blockage known as "soft gate blockage": the blockage is mainly composed of flexible materials such as wet wipes, fibrous fabrics, or plastic films, often adhering to pipe joints or tees. These materials exhibit typical nonlinear resistivity characteristics: when the inflow rate increases and the kinetic energy strengthens, the flexible material is washed away by the water flow, and the pipeline flow cross-section instantly recovers, giving the monitoring data the illusion of a sudden drop in water level and a surge in flow; however, when the flow rate decreases and the impact force weakens, the flexible material, affected by gravity or weak backflow, re-adheres to the pipe wall or blocks the flow cross-section, leading to a surge in local resistance and a renewed rise in water level.

[0005] This dynamic, fluctuating pattern can be misinterpreted as sensor malfunctions or transient hydraulic fluctuations in the current management logic, leading to frequent false alarms or missed alarms when maintenance is most needed. Because the existing system lacks in-depth analysis of the nonlinear hysteresis relationship between flow rate and level, and cannot capture physical characteristics such as soft material flutter, maintenance departments cannot accurately identify these highly concealed flexible blockages. Consequently, it is difficult to develop targeted preventative dredging plans, severely impacting the accuracy of asset status assessment and resource scheduling efficiency in pipeline network management. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide an intelligent underground drainage network management method and system to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A smart underground drainage network management method includes:

[0009] Acquire monitoring data from upstream monitoring points of the target pipe section, including real-time liquid level and real-time flow rate collected in a time series; based on the monitoring data, construct a liquid level-flow rate mapping model for the target pipe section, and divide the monitoring data into data for rising water periods and data for receding water periods;

[0010] A bidirectional resistance characteristic analysis is performed on the liquid level-flow mapping model, including identifying resistance release characteristics and resistance recovery characteristics. The resistance release characteristic is manifested as follows: during the rising water period or at the boundary between the rising water period and the receding water period, when the real-time liquid level reaches a first threshold, the real-time flow rate surges and the real-time liquid level experiences a sudden change in its rise or fall. The resistance recovery characteristic is manifested as follows: during the receding water period, when the real-time flow rate decreases to a second threshold, the real-time liquid level rebounds.

[0011] When the resistance release feature and the resistance recovery feature are detected simultaneously, it is preliminarily determined that there is a soft intermittent blockage anomaly in the target pipe section. The soft intermittent blockage anomaly indicates that there is a flexible obstruction in the target pipe section that repeatedly opens and closes with the change of water flow impact force.

[0012] Based on the abnormal result of soft intermittent blockage, the pipeline network operation and maintenance management process is executed, including generating a flexible dredging management work order for the target pipeline segment in the operation and maintenance work order management module, and updating the asset status marker in the pipeline network operation and maintenance database.

[0013] Preferably, the criterion for determining a surge in real-time traffic is: the first derivative of the real-time traffic with respect to time is greater than a preset traffic change rate threshold and the duration is not less than a preset threshold.

[0014] The criteria for determining a sudden change in the real-time liquid level, such as a decrease or a halt in the rise, are as follows: the first derivative of the real-time liquid level with respect to time changes from a positive value to a negative value and the duration is not less than a preset threshold, or the absolute value of the first derivative of the real-time liquid level with respect to time is less than a preset halt threshold and the duration is not less than a preset threshold.

[0015] The criterion for determining whether the real-time liquid level rebounds is: the first derivative of the real-time liquid level with respect to time changes from a negative value to a positive value and the duration is not less than a preset threshold.

[0016] Preferably, the upstream monitoring point is located in the inspection well at the upstream end of the target pipe section, the real-time liquid level is obtained by an ultrasonic level gauge installed at the top of the inspection well, and the real-time flow rate is obtained by multiplying the flow velocity measured by a Doppler velocity meter installed at the bottom of the inspection well with the flow cross-sectional area calculated based on the real-time liquid level.

[0017] Preferably, the first threshold corresponds to the opening liquid level value of the flexible barrier, the second threshold corresponds to the closing flow rate value of the flexible barrier, and the first threshold is higher than the liquid level height corresponding to the second threshold. The closed area between the two is defined as the hysteresis line area.

[0018] Determining that the target pipe segment has a soft intermittent blockage abnormality also includes secondary confirmation: calculating the area of ​​the hysteresis line region, and confirming the abnormality a second time when the area is greater than a preset hysteresis threshold.

[0019] Preferably, the method further includes intelligent screening based on machine learning:

[0020] A feature sample library for the target pipe section is constructed by using normal flow data and flexible obstruction blockage data identified from historical monitoring data;

[0021] Extract the geometric contour features of the liquid level and flow rate correlation curves from the feature sample library, and use a classification training algorithm to establish a classification and identification model for soft intermittent blockage anomalies;

[0022] The trajectory features in the real-time constructed liquid level-flow mapping model are input into the classification and recognition model;

[0023] When the identification result output by the classification and recognition model matches the soft intermittent blockage anomaly, and its corresponding confidence score exceeds the preset confidence threshold, the step of generating a flexible dredging management work order is executed.

[0024] Preferably, the method further includes a flutter detection step based on pressure signals:

[0025] Acquire pressure data collected by pressure sensors at upstream monitoring points of the target pipe section;

[0026] During the resistance release characteristic and the resistance recovery characteristic, the spectral characteristics of the pressure data are analyzed;

[0027] If an energy peak within a preset frequency range is detected based on the spectral characteristics, the flexible barrier is determined to be in an open and fluttering state, and this is used as a secondary basis for determining the soft intermittent blockage anomaly.

[0028] Preferably, the method further includes an audio signal-based flutter detection step:

[0029] Acquire audio data collected by the audio sensor at the upstream monitoring point of the target pipe section;

[0030] During the resistance release and resistance recovery phases, the spectral characteristics of the audio data are analyzed.

[0031] If an energy peak within a preset frequency range is detected based on the spectral characteristics, the flexible barrier is determined to be in an open and fluttering state, and this is used as a secondary basis for determining the soft intermittent blockage anomaly.

[0032] Preferably, the preset frequency range is 100Hz to 2000Hz.

[0033] Preferably, updating the asset status markers in the pipeline network operation and maintenance database specifically includes:

[0034] The target pipe segment is marked as a potential soft intermittent blockage anomaly, and a resistance characteristic profile of the target pipe segment is established, including recording historical change data of the first threshold and the second threshold;

[0035] The hardening trend of the flexible barrier is analyzed based on the historical change data.

[0036] When the first threshold shows an upward trend, it is determined that the flexible barrier is fibrous or calcified, and the treatment priority of the target pipe segment is increased.

[0037] This invention also discloses an intelligent underground drainage network management system for implementing the above method, comprising:

[0038] The monitoring data acquisition module is used to acquire monitoring data from upstream monitoring points of the target pipe section. The monitoring data includes real-time liquid level and real-time flow rate collected according to a time series.

[0039] The data processing and modeling module is used to construct a liquid level-flow mapping model for the target pipe section based on the monitoring data, and to divide the monitoring data into data during the rising water period and data during the receding water period.

[0040] The resistance feature analysis module is used to perform bidirectional resistance feature analysis on the liquid level-flow mapping relationship model, and to identify resistance release features and resistance recovery features respectively.

[0041] The resistance release characteristic is manifested as follows: during the rising water period or at the boundary between the rising water period and the receding water period, when the real-time liquid level reaches the first threshold, the real-time flow rate surges and the real-time liquid level drops or stops rising abruptly; the resistance recovery characteristic is manifested as follows: during the receding water period, when the real-time flow rate decreases to the second threshold, the real-time liquid level rebounds.

[0042] An anomaly detection module is used to initially determine that there is a soft intermittent blockage anomaly in the target pipe section when the resistance release feature and the resistance recovery feature are detected simultaneously. The soft intermittent blockage anomaly indicates that there is a flexible obstruction in the target pipe section that repeatedly opens and closes with the change of water flow impact force.

[0043] The management and maintenance module is used to generate a flexible dredging management work order for the target pipe section based on the determination result of the soft intermittent blockage anomaly, and update the asset status marker in the pipeline network operation and maintenance database.

[0044] The advantage of this invention over existing technologies lies in its core principle: utilizing the hydraulic resistance hysteresis law exhibited by underground drainage pipe sections when flexible obstructions are present. When the incoming water volume increases, the flexible obstruction opens under the influence of head / impact force, causing a sudden decrease in the equivalent resistance of the pipe section. Conversely, when the incoming water volume decreases, it closes under its own weight, rebound, or adhesion force, causing the equivalent resistance of the pipe section to increase again. This opening-closing reciprocating process manifests as a pair of resistance release characteristics during the rising water phase and resistance recovery characteristics during the receding water phase in the time sequence of liquid level and flow rate at the same monitoring point. Furthermore, these two characteristics coincide with the switching of hydraulic states in time, distinguishing them from the slow changes caused by general siltation or occasional anomalies caused by single hydraulic fluctuations.

[0045] Based on the above patterns, this invention constructs a level-flow mapping model by analyzing the real-time liquid level and flow rate at upstream monitoring points. The monitoring data is divided into data for rising and receding water periods, and further bidirectional resistance characteristic analysis is conducted. During rising water periods, the resistance release criterion is "the flow rate surges and the level rise stagnates or decreases when the real-time liquid level reaches the first threshold." During receding water periods, the resistance recovery criterion is "the liquid level rebounds when the real-time flow rate decreases to the second threshold." When both types of characteristics are detected simultaneously, "paired hysteresis features" can be used as the basis for determining soft intermittent blockage anomalies. This avoids misjudging such intermittent and nonlinear anomalies as sensor noise, occasional hydraulic fluctuations, or ordinary slow siltation. Furthermore, anomaly determination can directly drive the operation and maintenance management process, automatically generating flexible dredging management work orders and updating the asset status markers in the pipeline network operation and maintenance database, improving the timeliness of handling and the efficiency of closed-loop management.

[0046] Furthermore, this invention, by corresponding the first threshold and the second threshold to the opening liquid level value and closing flow rate value of the flexible barrier respectively, and by introducing the hysteresis line region and its area as the basis for secondary confirmation, makes the judgment not only dependent on a single feature trigger, but also verifiable through hysteresis intensity quantification, thereby improving the robustness to "sometimes open and sometimes closed" anomalies and reducing the probability of false dispatch.

[0047] This invention also constructs a feature sample library of normal flow and flexible obstruction blockage, extracts the geometric contour features of the liquid level-flow correlation curve and trains a classification and recognition model, and filters it with confidence threshold after real-time trajectory input. It can achieve adaptive recognition under different pipe sections, different rainfall patterns and different baseline conditions, enhance the cross-scenario generalization ability and further reduce misjudgment and missed judgment under boundary conditions.

[0048] In addition, by introducing spectral analysis of pressure or audio signals during resistance release and resistance recovery, and detecting energy peaks within a preset frequency range to identify the "on and fluttering state," a secondary judgment basis is formed that corroborates the liquid level-flow characteristics, improving the observability of the mechanistic phenomenon of flutter in flexible components. This enhances the reliability of the judgment when hydraulic fluctuations, backflow, or local disturbances are present. The preset frequency range is limited to 100Hz to 2000Hz, which helps to focus the detection on the frequency band where typical flutter sound / pressure energy is more concentrated, reducing confusion with low-frequency background fluctuations.

[0049] Furthermore, by establishing resistance characteristic files in the operation and maintenance database and recording the historical changes of the first and second thresholds, trend analysis of the state evolution of flexible barriers can be performed. When the first threshold shows an upward trend, the priority of handling can be increased, and abnormal handling can be extended to asset health management, realizing the upgrade from one-time alarm to predictive operation and maintenance, thereby more rationally allocating dredging resources and reducing the risk of sudden overflow.

[0050] In summary, this invention is based on the resistance hysteresis law caused by flexible obstructions, and transforms the observable characteristics of hydraulic mechanisms into a decision-making basis for pipeline operation and maintenance management that can be implemented in an engineering manner, verified, and executed in a closed loop. This not only improves the accuracy and stability of soft intermittent blockage identification, but also strengthens the closed-loop management capability of work orders and asset ledgers. Attached Figure Description

[0051] Figure 1 This is a flowchart of the method of the present invention;

[0052] Figure 2 This is a graph of the liquid level-flow rate mapping model of the present invention;

[0053] Figure 3 This is a schematic diagram showing the three states of the flexible barrier of the present invention;

[0054] Figure 4 This is a schematic diagram of the high-frequency vibration recognition method of the present invention;

[0055] Figure 5 This is a schematic diagram of the system of the present invention. Detailed Implementation

[0056] The specific embodiments of the present invention will now be described with reference to the accompanying drawings.

[0057] Example 1:

[0058] This embodiment provides a smart underground drainage network management method. Traditional network monitoring often struggles to distinguish between permanent hard deposits such as silt and intermittent soft blockages such as flexible floating objects like woven bags and plastic films. The characteristic of flexible blockages is that their obstruction of water flow is non-linear and varies with hydrodynamic conditions. To address this challenge, the core logic of this embodiment lies in utilizing the hysteresis effect of liquid level and flow rate over time to capture the breathing characteristics of flexible objects.

[0059] like Figure 1 As shown, the method of this invention first requires acquiring monitoring data from an upstream monitoring point of the target pipe section. The upstream monitoring point is chosen because when a blockage occurs in the pipe section, the backflow phenomenon is first and most significantly manifested upstream of the blockage point. The monitoring data includes real-time liquid level and real-time flow rate collected in a time series. In terms of hardware deployment, the equipment can be installed in a manhole at the upstream end of the target pipe section. Specifically, the real-time liquid level can be obtained using an ultrasonic level gauge installed at the top of the manhole, a location that avoids debris entanglement; the real-time flow rate is obtained by multiplying the flow velocity measured by a Doppler velocimeter installed at the bottom of the manhole by the flow cross-sectional area calculated based on the real-time liquid level.

[0060] After acquiring the data, the system constructs a level-flow mapping model (HQ relationship) for the target pipe section based on the monitoring data. To capture dynamic characteristics, the system divides the monitoring data into rising water period data and receding water period data. In one embodiment, the data division is directly determined based on the trend of the real-time level curve obtained from the target monitoring point. Specifically, the first derivative of the real-time level with respect to time can be calculated. The time window in which the level continues to rise and the derivative value is greater than zero is marked as the rising water period, while the time window in which the level continues to fall and the derivative value is less than zero is marked as the receding water period. This method relies entirely on local sensor data, has simple calculation logic, and can directly reflect the physical state of the monitoring section. In another embodiment, considering the possibility of local fluctuations within the pipe network, the division criteria can be further combined with the water conditions at the upstream end of the converging pipe. Specifically, it can be determined by obtaining the rain gauge data associated with the drainage area or the operating status of the upstream associated pumping station. For example, the period from the start of rainfall to the peak of rainfall, or the stage when the upstream pumping station starts and supplies water to the pipe section, can be defined as the rising water period, while the decay period after the rainfall stops or the emptying period after the pumping station is shut down can be defined as the receding water period. This boundary condition-based division method can divide the flow process from the hydraulic source in a more physical way, which helps to more accurately capture the complete resistance change cycle in complex flow fields.

[0061] Subsequently, the system performs a two-way resistance characteristic analysis on the liquid level-flow mapping model. For example... Figure 2 As shown, this analysis aims to identify resistance release and resistance recovery characteristics. This design is based on the physical properties of the flexible barrier: when the upstream water pressure increases, the flexible object is pushed aside (resistance release); when the water flow weakens, the object re-covers the pipe opening under its own weight or elastic action (resistance recovery).

[0062] The specific identification logic is as follows: Resistance release characteristics are manifested as follows: During periods of rising water or at the boundary between rising and receding water, when the real-time liquid level reaches the first threshold, the real-time flow surges and the real-time liquid level experiences a sudden change, either stagnating or declining. Physically, this means that the accumulated water head finally breaks through the flexible barrier, causing the accumulated water to drain rapidly (increased flow, decreased liquid level). Resistance recovery characteristics are manifested as follows: During periods of receding water, when the real-time flow drops to the second threshold, the real-time liquid level rebounds. This indicates that as the scouring force weakens, the barrier closes again, causing water to accumulate upstream again. The first threshold is the opening liquid level threshold, used to characterize the triggering liquid level for the flexible barrier to transition from a closed to an open state. Physically, it means that the head difference and impact generated after the upstream water level rises to a certain height reach the critical condition required for the flexible barrier to be opened. Therefore, when the real-time liquid level reaches the first threshold, a sudden increase in water flow capacity is more likely to occur, accompanied by a stagnation or decline in the liquid level rise. The second threshold is the closing flow threshold, which is used to characterize the trigger flow rate for the flexible barrier to return from the open state to the closed state. Its physical meaning is that when the flow rate drops to a certain level during the receding process, the supporting and pulling effect of the water flow on the flexible barrier is insufficient to maintain its opening. The flexible barrier closes under its own weight, rebound or adhesion, thereby increasing the equivalent resistance and causing the upstream liquid level to rebound. The two thresholds correspond to different triggering conditions for opening and closing, forming hysteresis characteristics, which makes the same pipe section exhibit different liquid level-flow trajectories during the rising and receding stages.

[0063] When both resistance release and resistance recovery characteristics are detected simultaneously, the system initially determines that the target pipe section has a soft, intermittent blockage anomaly. This anomaly indicates the presence of a flexible obstruction within the pipe section that repeatedly opens and closes in response to changes in water flow impact force, such as... Figure 3 As shown, this illustrates the switching of a flexible barrier from closed to open and back to closed.

[0064] Finally, based on the abnormal results of soft intermittent blockage, the pipeline network operation and maintenance management process is executed, including generating flexible dredging management work orders for the target pipeline segment in the operation and maintenance work order management module, and updating the asset status markers in the pipeline network operation and maintenance database, thereby achieving precise operation and maintenance.

[0065] Example 2:

[0066] Based on Example 1, this embodiment performs specific mathematical quantification on the criteria for determining mutations and thresholds to eliminate interference from sensor noise and conventional hydraulic fluctuations.

[0067] This embodiment employs derivative analysis to determine when a surge in real-time flow occurs. The determination criterion is set as follows: the first derivative of the real-time flow with respect to time is greater than a preset flow change rate threshold, and the duration is not less than a preset threshold (selected as 1 to 3 minutes). The flow change rate threshold should be selected with reference to the maximum slope of flow increase in historical normal rainfall events, typically 1.5 to 2 times that maximum slope, for example, set to 10 (cubic meters / hour) / minute, to ensure that the identified surge is significantly different from normal confluence.

[0068] For sudden changes in real-time liquid level, such as a drop or stagnation, the judgment criteria are divided into two cases: First, the first derivative of the real-time liquid level with respect to time changes from a positive value to a negative value for a duration not less than a preset threshold (selected as 1 to 3 minutes), which corresponds to a significant surge followed by a drop; second, the absolute value of the first derivative of the real-time liquid level with respect to time is less than a preset stagnation threshold for a duration not less than a preset threshold (selected as 3 to 5 minutes), which corresponds to stagnation. The stagnation threshold can be taken as 0.5 mm / min to 2 mm / min.

[0069] The criteria for determining whether the real-time liquid level rebounds are: the first derivative of the real-time liquid level with respect to time changes from a negative value to a positive value and the duration is not less than a preset threshold (selected as 3 to 5 minutes).

[0070] Furthermore, to further improve the accuracy of the determination, this embodiment introduces the hysteresis region area as a secondary confirmation criterion. As mentioned above, the first threshold corresponds to the opening liquid level value of the flexible barrier, and the second threshold corresponds to the closing flow rate value of the flexible barrier. Physically, the opening liquid level is usually higher than the maintaining liquid level when closed. The closed region between the two is defined as the hysteresis region, such as... Figure 2 The area enclosed by the hysteresis curve. The specific steps for secondary confirmation are: calculate the area of ​​the hysteresis region; if the area is greater than a preset hysteresis threshold, the secondary confirmation is abnormal. The area of ​​the hysteresis region can be calculated using existing methods for calculating the area inside a closed curve, where the closed curve region is the closed area between the first and second thresholds of the data curve collected in the liquid level-flow coordinate system of this invention. The area value can be obtained by using the trapezoidal approximation method in existing technology within the closed region, that is, dividing the closed region into multiple approximate trapezoids along the longitudinal or transverse direction, and calculating the area by adding the areas of the trapezoids. This area calculation can be implemented using existing numerical analysis software on a computer.

[0071] In one embodiment, the hysteresis threshold is three times the normal baseline value. This is because the hysteresis loop area generated by abnormal flexible obstruction is usually much larger than that generated under normal operating conditions. In some embodiments, if the calculated area exceeds three times the normal baseline value, it is confirmed as a soft blockage. If a closed curve is not formed under normal operating conditions, in a specific embodiment, in order to ensure that the liquid level-flow trajectory under normal operating conditions can also form an area diameter consistent with the hysteresis loop of the present invention, the closed region is obtained using the same coordinate cutting method as the abnormal hysteresis loop. Specifically, under abnormal operating conditions, the horizontal or vertical coordinate range of the hysteresis loop is first determined, and this range is used as the cutting boundary. Under normal operating conditions, the normal liquid level-flow trajectory is restricted to the same horizontal coordinate range, and two curves are taken for the rising and falling sections within this range. The closed area enclosed by the two curves and the horizontal coordinate boundary is taken as the normal closed area. Alternatively, the normal liquid level-flow trajectory is restricted to the same vertical coordinate range, and two curves are taken for the rising and falling sections within this range. The closed area enclosed by the two curves and the vertical coordinate boundary is taken as the normal closed area. The area of ​​the closed area calculated in this way can be taken as the average value of the closed area measured multiple times as the above-mentioned normal benchmark value, which is used to compare and determine with the area of ​​the abnormal hysteresis loop.

[0072] Example 3:

[0073] This embodiment further introduces a machine learning-based intelligent screening mechanism to handle complex and ever-changing pipeline environments and solve the problem of misjudgment in critical situations using the traditional threshold method.

[0074] The process mainly includes three stages: feature sample library construction, model training, and real-time identification. First, using manually verified records from historical monitoring data, normal flow data is identified as negative samples and data blocked by flexible obstructions as positive samples to construct a feature sample library for the target pipe section. Second, the geometric contour features of the liquid level-flow correlation curve are extracted from the feature sample library. The selected features include:

[0075] First, the total length of the hysteresis loop. Each discrete sampling point on the hysteresis loop includes the flow rate and liquid level values ​​at the corresponding time. The cumulative value obtained by summing the planar distances between adjacent points is used as the total length of the hysteresis loop.

[0076] Second, the amplitude range of the hysteresis curve. The maximum and minimum flow rates along the flow axis and the maximum and minimum liquid levels along the level axis are statistically analyzed. The differences between the maximum and minimum flow rates and the differences between the maximum and minimum liquid levels are used as the corresponding amplitude ranges to characterize the span of the curve in the flow and level directions.

[0077] Third, the area of ​​the hysteresis region can be calculated using the trapezoidal approximation method.

[0078] A classification and identification model for soft intermittent blockage anomalies is established using a classification training algorithm. In this embodiment, the model architecture employs a support vector machine or a random forest classifier. If a random forest model is used, the number of decision trees is recommended to be between 100 and 200, and the maximum depth is controlled between 10 and 15 layers to prevent overfitting. During model training, cross-validation is used to adjust hyperparameters, and grid search is used to determine the optimal classification boundary. Finally, the trajectory features from the real-time constructed level-flow mapping model are input into the trained classification and identification model. Only when the identification result output by the classification and identification model matches a soft intermittent blockage anomaly, and its corresponding confidence score exceeds a preset confidence threshold, such as 0.85 or 85%, is the step of generating a flexible dredging management work order executed. This method integrates the advantages of data-driven approaches and can effectively identify complex hysteresis curves with irregular shapes.

[0079] In another embodiment, intelligent screening can also employ an end-to-end temporal neural network. The network structure uses 1D convolutions plus bidirectional gated recurrent units, with 3 convolutional layers, 5 to 15 kernels, and 32, 64, and 128 channels respectively, and a dropout rate of 0.1 to 0.3. The recurrent units have 64 to 256 hidden units and 1 or 2 layers. After time-dimensional aggregation, a fully connected layer is connected, and p is output through a sigmoid function. Training samples are obtained from historical data slices, and abnormal samples come from work order closed-loop confirmation or random inspection confirmation segments. The loss function is weighted binary cross-entropy, with abnormal class weights of 2 to 10. The optimizer is Adam, with a learning rate of 0.0001 to 0.001, a batch size of 32 to 256, 20 to 200 training epochs, and an early stopping strategy. During online judgment, p is compared with a confidence threshold; if it exceeds the threshold, a work order is generated. If pressure or audio data is collected simultaneously, the spectral energy sequence of a preset frequency band can be used as an additional input channel to enhance boundary condition discrimination.

[0080] Example 4:

[0081] This embodiment focuses on using the frequency domain characteristics of physical signals for auxiliary verification. Flexible objects often produce high-frequency flutter under the scouring of water flow, which is a significant characteristic that distinguishes them from rigid sediment.

[0082] like Figure 4As shown, this embodiment configures a high-frequency pressure sensor or an audio sensor at the monitoring point. The method includes a flutter detection step based on pressure signals or a flutter detection step based on audio signals. The system acquires pressure data or audio data collected from monitoring points upstream of the target pipe section. During the period when the aforementioned resistance release and resistance recovery characteristics are detected (i.e., within the suspected opening and closing time window), a short-time Fourier transform is performed on the data to analyze the spectral characteristics. If an energy peak within a preset frequency range is detected based on the spectral characteristics, it is determined that the flexible blockage is in an open and fluttering state. The preset frequency range is selected as 100Hz to 2000Hz. The reason for selecting this frequency band is that the self-excited vibration frequency of common underground pipe network flexible waste such as woven bags and raincoats under water flow impact is mainly concentrated in this range, while background water flow noise is usually concentrated in the low-frequency band below 100Hz. Detecting the energy peak in this frequency band can serve as a strong secondary basis for determining soft intermittent blockage anomalies.

[0083] Example 5:

[0084] This embodiment focuses on the full lifecycle management and predictive maintenance of assets.

[0085] When updating the asset status tags in the pipeline network operation and maintenance database, the system not only records current anomalies but also performs long-term trend analysis. Specifically, this includes: marking target pipe segments as potential soft intermittent blockage anomalies and establishing a resistance characteristic profile for the target pipe segments. The profile focuses on recording historical changes in the first threshold (open liquid level) and the second threshold (closed flow rate).

[0086] Based on historical data, the system analyzes the hardening trend of flexible barriers. In actual operating conditions, flexible objects undergo fibrosis, calcification, or adsorption hardening after prolonged immersion, leading to a decrease in their flexibility and a gradual increase in the opening pressure required to be dislodged by water flow. Therefore, when an upward trend in the first threshold is detected, for example, an average monthly increase of 5 cm in the opening liquid level over the past three months, it is determined that the flexible barrier is undergoing fibrosis or calcification. At this point, the system automatically prioritizes the treatment of the target pipe section, alerting maintenance personnel that the risk of blockage at that location is escalating and may soon develop into a complete blockage, requiring priority cleaning.

[0087] In one embodiment, the cleaning method corresponding to the flexible dredging management work order is a flexible cleaning method, which means that without using high-rigidity blades to forcibly cut, the reversible deformation characteristics of flexible obstructions are utilized for treatment. Specifically, this may include using high-pressure pulse flushing water to directionally flush the target pipe section, causing the flexible obstruction to be flushed away from the attached silt and recover its elasticity when it is open; or using an air-water pulse dredging method, injecting compressed air and water into the pipe at the inspection well to form alternating pulses, so as to induce the flexible obstruction to open periodically and carry away the entangled material; or using a soft traction cleaning method, using soft ropes or flexible straps to pull back and forth between inspection wells to pull out fibrous debris stuck on the edge of the flexible obstruction. After the cleaning is completed, the liquid level and flow rate data of another round of water rise and fall can be collected again to verify whether the resistance release and resistance recovery characteristics have disappeared and whether the hysteresis area area has fallen back, thereby forming a closed-loop record of the work order and updating the asset status mark.

[0088] Example 6:

[0089] This embodiment provides an intelligent underground drainage network management system for implementing the above-described method. For example... Figure 5 As shown, the system includes a monitoring data acquisition module, a data processing and modeling module, a resistance characteristic analysis module, an anomaly detection module, and a management and maintenance module.

[0090] The monitoring data acquisition module connects to on-site sensors to obtain real-time liquid level and flow rate. The data processing and modeling module is used for data cleaning, segmentation (flooding / receding), and building an HQ model. The resistance characteristic analysis module is the core computing unit of the system, with built-in algorithms for identifying resistance release and resistance recovery, specifically executing the threshold determination and derivative calculation in Examples 1 and 2. The anomaly determination module combines logical rule determination (Example 1) with optional machine learning models (Example 3) and spectrum analysis (Example 4) to output the final anomaly conclusion. The management and maintenance module is responsible for connecting to the enterprise's work order system, generating flexible dredging management work orders based on the anomaly determination results, and updating asset status and disposal priority according to the logic described in Example 5.

[0091] By combining the above embodiments, the present invention can effectively solve the problem of soft intermittent blockage in drainage pipe networks, which is highly concealed and difficult to identify, and improve the operational safety and maintenance efficiency of urban drainage systems.

[0092] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A smart underground drainage network management method, characterized in that, include: Acquire monitoring data from upstream monitoring points of the target pipe section, including real-time liquid level and real-time flow rate collected in time series; Based on the monitoring data, a liquid level-flow mapping model for the target pipe section is constructed, and the monitoring data is divided into data during the rising water period and data during the receding water period. A bidirectional resistance characteristic analysis is performed on the liquid level-flow mapping model, including identifying resistance release characteristics and resistance recovery characteristics. The resistance release characteristic is manifested as follows: during the rising water period or at the boundary between the rising water period and the receding water period, when the real-time liquid level reaches a first threshold, the real-time flow rate surges and the real-time liquid level experiences a sudden change in its rise or fall. The resistance recovery characteristic is manifested as follows: during the receding water period, when the real-time flow rate decreases to a second threshold, the real-time liquid level rebounds. When the resistance release feature and the resistance recovery feature are detected simultaneously, it is preliminarily determined that there is a soft intermittent blockage anomaly in the target pipe section. The soft intermittent blockage anomaly indicates that there is a flexible obstruction in the target pipe section that repeatedly opens and closes with the change of water flow impact force. Based on the abnormal result of soft intermittent blockage, the pipeline network operation and maintenance management process is executed, including generating a flexible dredging management work order for the target pipeline segment in the operation and maintenance work order management module, and updating the asset status marker in the pipeline network operation and maintenance database.

2. The method according to claim 1, characterized in that: The criteria for determining a surge in real-time traffic are: the first derivative of the real-time traffic with respect to time is greater than a preset traffic change rate threshold and the duration is not less than a preset threshold. The criteria for determining a sudden change in the real-time liquid level, such as a decrease or a halt in the rise, are as follows: the first derivative of the real-time liquid level with respect to time changes from a positive value to a negative value and the duration is not less than a preset threshold, or the absolute value of the first derivative of the real-time liquid level with respect to time is less than a preset halt threshold and the duration is not less than a preset threshold. The criterion for determining whether the real-time liquid level rebounds is: the first derivative of the real-time liquid level with respect to time changes from a negative value to a positive value and the duration is not less than a preset threshold.

3. The method according to claim 1, characterized in that, The upstream monitoring point is set in the inspection well at the upstream end of the target pipe section. The real-time liquid level is obtained by an ultrasonic level gauge installed at the top of the inspection well. The real-time flow rate is obtained by multiplying the flow velocity measured by a Doppler velocity meter installed at the bottom of the inspection well with the cross-sectional area of ​​the flow calculated based on the real-time liquid level.

4. The method according to claim 1, characterized in that, The first threshold corresponds to the opening liquid level value of the flexible barrier, the second threshold corresponds to the closing flow rate value of the flexible barrier, and the first threshold is higher than the liquid level height corresponding to the second threshold. The closed area between the two is defined as the hysteresis line area. Determining that the target pipe segment has a soft intermittent blockage abnormality also includes secondary confirmation: calculating the area of ​​the hysteresis line region, and confirming the abnormality a second time when the area is greater than a preset hysteresis threshold.

5. The intelligent underground drainage network management method according to claim 1, characterized in that, The method further includes intelligent screening based on machine learning: A feature sample library for the target pipe section is constructed by using normal flow data and flexible obstruction blockage data identified from historical monitoring data; Extract the geometric contour features of the liquid level and flow rate correlation curves from the feature sample library, and use a classification training algorithm to establish a classification and identification model for soft intermittent blockage anomalies; The trajectory features in the real-time constructed liquid level-flow mapping model are input into the classification and recognition model; When the identification result output by the classification and recognition model matches the soft intermittent blockage anomaly, and its corresponding confidence score exceeds the preset confidence threshold, the step of generating a flexible dredging management work order is executed.

6. The method according to claim 1, characterized in that, The method also includes a flutter detection step based on pressure signals: Acquire pressure data collected by pressure sensors at upstream monitoring points of the target pipe section; During the resistance release characteristic and the resistance recovery characteristic, the spectral characteristics of the pressure data are analyzed; If an energy peak within a preset frequency range is detected based on the spectral characteristics, the flexible barrier is determined to be in an open and fluttering state, and this is used as a secondary basis for determining the soft intermittent blockage anomaly.

7. The method according to claim 1, characterized in that, The method also includes a flutter detection step based on audio signals: Acquire audio data collected by the audio sensor at the upstream monitoring point of the target pipe section; During the resistance release and resistance recovery phases, the spectral characteristics of the audio data are analyzed. If an energy peak within a preset frequency range is detected based on the spectral characteristics, the flexible barrier is determined to be in an open and fluttering state, and this is used as a secondary basis for determining the soft intermittent blockage anomaly.

8. The method according to claim 6 or 7, characterized in that, The preset frequency range is 100Hz to 2000Hz.

9. The method according to claim 1, characterized in that, Updating the asset status markers in the pipeline network operation and maintenance database specifically includes: The target pipe segment is marked as a potential soft intermittent blockage anomaly, and a resistance characteristic profile of the target pipe segment is established, including recording historical change data of the first threshold and the second threshold; The hardening trend of the flexible barrier is analyzed based on the historical change data. When the first threshold shows an upward trend, it is determined that the flexible barrier is fibrous or calcified, and the treatment priority of the target pipe segment is increased.

10. A smart underground drainage network management system for implementing the method of claim 1, characterized in that, include: The monitoring data acquisition module is used to acquire monitoring data from upstream monitoring points of the target pipe section. The monitoring data includes real-time liquid level and real-time flow rate collected according to a time series. The data processing and modeling module is used to construct a liquid level-flow mapping model for the target pipe section based on the monitoring data, and to divide the monitoring data into data during the rising water period and data during the receding water period. The resistance feature analysis module is used to perform bidirectional resistance feature analysis on the liquid level-flow mapping relationship model, and to identify resistance release features and resistance recovery features respectively. The resistance release characteristic is manifested as follows: during the rising water period or at the boundary between the rising water period and the receding water period, when the real-time liquid level reaches the first threshold, the real-time flow rate surges and the real-time liquid level drops or stops rising abruptly; the resistance recovery characteristic is manifested as follows: during the receding water period, when the real-time flow rate decreases to the second threshold, the real-time liquid level rebounds. An anomaly detection module is used to initially determine that there is a soft intermittent blockage anomaly in the target pipe section when the resistance release feature and the resistance recovery feature are detected simultaneously. The soft intermittent blockage anomaly indicates that there is a flexible obstruction in the target pipe section that repeatedly opens and closes with the change of water flow impact force. The management and maintenance module is used to generate a flexible dredging management work order for the target pipe section based on the determination result of the soft intermittent blockage anomaly, and update the asset status marker in the pipeline network operation and maintenance database.

Images