Early warning method and system for pipe explosion of underground comprehensive pipe gallery

Abstract

The invention discloses an underground comprehensive pipe gallery pipe explosion early warning method and system, and relates to the field of safety monitoring of underground comprehensive pipe galleries in urban infrastructure construction. According to the method, a high-frequency pressure meter, a hydrophone and a flow meter are installed at key positions of a water supply pipeline, pipeline pressure, sound and flow data are monitored in real time, pressure transient change, abnormal noise and flow change in the pipeline are recognized through a signal analysis algorithm and a machine learning algorithm, and therefore early warning of a pipe explosion event is achieved. And the operation safety and reliability of the pipe gallery are improved.

Description

Technical Field

[0001] The present invention relates to a safety monitoring technology for underground integrated pipe corridors in urban infrastructure construction, and specifically to an underground integrated pipe corridor pipe burst early warning method and system, in particular to an underground integrated pipe corridor pipe burst early warning method and system based on a high-frequency pressure gauge, a hydrophone and a flow meter. Background Art

[0002] With the acceleration of urbanization, the safety of underground integrated pipe corridors, as an important part of urban infrastructure, has received more and more attention in terms of construction and operation. As an important part of the pipe corridor, once a pipe burst occurs, the water supply pipeline will not only affect the normal operation of the water supply, but also may cause damage to other pipelines and facilities in the pipe corridor, and even cause safety accidents. Therefore, the development of an effective pipe burst warning system is of great significance to ensure the safe operation of the pipe corridor. Summary of the invention

[0003] In view of the problem that the existing technology of underground integrated pipe gallery water supply pipe burst warning capability is insufficient, the present invention proposes an underground integrated pipe gallery pipe burst warning method and system. The method realizes early warning of pipe burst events by real-time monitoring of pipeline pressure, sound and flow data, combined with advanced signal analysis algorithms and machine learning algorithms, and reduces the impact of emergencies on pipe gallery and urban infrastructure.

[0004] Technical solution: An underground integrated pipe gallery pipe burst warning system, which includes a data acquisition unit, a data transmission unit, a data analysis unit and a warning response unit. It can monitor and analyze the key parameters of the water supply pipeline in real time, accurately identify potential pipe burst risks through signal processing and machine learning algorithms, and issue warning signals in a timely manner;

[0005] Data acquisition unit: including high-frequency pressure gauge, hydrophone and flow meter, real-time monitoring of pipeline pressure, sound and flow data;

[0006] Data transmission unit: used to transmit the data collected by the data collection unit to the data analysis unit, and the data transmission unit includes wired connection or wireless transmission;

[0007] Data analysis unit: Receives data from the data acquisition unit through the data transmission unit, and processes the pressure, sound and flow data through the signal analysis algorithm, including obtaining characteristic parameters; also includes using the trained machine learning model to analyze the characteristic parameters to identify pressure transients, abnormal noise and flow changes in the pipeline, so as to determine whether there is a risk of pipe burst;

[0008] Early warning response unit: If the data analysis unit determines that there is a risk of pipe explosion, the early warning response unit will be activated immediately and send early warning information to the supervisor.

[0009] In the system, the data analysis unit includes the following data processing:

[0010] (1) Pressure data

[0011] Pressure change magnitude: Calculate the difference between adjacent data points to obtain the magnitude of the pressure change:

[0012] DeltaP=P(t i+1 )-P(t i )

[0013] Among them, P(t i ) represents the time point t i The pressure value;

[0014] Pressure change frequency: The frequency distribution of pressure change is obtained by performing Fourier transform or fast Fourier transform on the pressure data:

[0015]

[0016] Among them, P(f) is the component of pressure at frequency f, and P(t) is the pressure signal in the time domain.

[0017] Duration of pressure change: record the length of time when the pressure exceeds a certain threshold, including extracting characteristic parameters such as the amplitude, frequency, and duration of pressure change; this characteristic parameter can reflect the change of pressure in the pipeline and help identify water hammer signals;

[0018] (2) Sound Data Characteristics

[0019] Spectrum analysis: By performing Fourier transform on the sound signal, the spectrum characteristics of the sound are obtained, so as to identify abnormal sounds including water flow and leakage.

[0020]

[0021] Among them, S(f) is the component of the sound at frequency f, and s(t) is the sound signal in the time domain;

[0022] Energy distribution: Calculate the energy distribution of the sound signal at each frequency to identify the sound characteristics:

[0023] E((f)=|S(f)| 2

[0024] Where E(f) is the sound energy at frequency f;

[0025] Extract the spectrum characteristics and energy distribution of the sound signal, and through spectrum analysis, identify abnormal water flow or leakage sounds, which are usually associated with pipe burst events;

[0026] (3) Traffic data characteristics

[0027] Flow rate change amplitude: Similar to the pressure change amplitude, calculate the difference between adjacent data points:

[0028] DeltaQ=Q(t i+1 )-Q(t i )

[0029] Among them, Q(t i ) represents the time point t i The flow value;

[0030] Flow rate of change: Calculate the derivative of the flow to obtain the flow rate change trend:

[0031]

[0032] By extracting characteristic parameters including the amplitude and frequency of flow changes, flow data can reflect the flow state of the fluid in the pipeline and help determine whether there are abnormal flow fluctuations.

[0033] Furthermore, the data acquisition unit includes a high-frequency pressure gauge, a hydrophone, and a flow meter;

[0034] The high-frequency pressure gauge and hydrophone are arranged at the turning point, branch point, vulnerable section, and the same point. The high-frequency pressure gauge collects the flow fluctuation caused by the opening of the valve, the start and stop of the water pump, and the water use of the user, as well as the water pressure fluctuation, and transmits the monitoring data to the data analysis unit; the hydrophone is used in conjunction with the high-frequency pressure gauge to identify whether there is a pipe burst or abnormal noise by collecting sound data in the pipe network. The abnormal noise includes water leakage sound and vibration data;

[0035] The flow meter is arranged at a non-turning or valve position to avoid areas of eddy current or turbulence in the corresponding pipe gallery.

[0036] Furthermore, the early warning response unit includes a remote control module to realize control of valves, and the pipe gallery includes one or more control valves;

[0037] The early warning response unit also includes sending early warning information to supervisors, and the sending of early warning information includes SMS, email, and APP push.

[0038] An underground integrated pipe gallery pipe burst early warning method, the method is executed by the above system, and comprises the following steps:

[0039] S1. Deploy data acquisition units on the water supply pipelines in the integrated pipe gallery, including high-frequency pressure gauges, hydrophones and flow meters, to monitor pipeline pressure, sound and flow data in real time, and pre-process the collected data, including cleaning and denoising, data compression and data conversion, so as to reduce the cost of data transmission and storage, and obtain data that can be used for signal analysis algorithm recognition;

[0040] S2. Use a combination of machine learning models and signal analysis algorithms to build a pipe burst risk prediction model. The construction of the model includes using statistical analysis and signal processing technology to perform abnormal analysis on the data collected from the water supply pipe, including feature extraction of the data collected by the data acquisition unit, and inputting the processed data into the trained machine learning model to obtain a pipe burst risk prediction model to identify possible abnormal patterns, including comparing the current pressure and flow changes with the historical data to determine whether there are abnormal pressure and flow fluctuations; and identifying abnormal water flow sounds or leakage sounds through spectrum analysis of sound signals;

[0041] S3, inputting the real-time collected and pre-processed data into the trained pipe burst risk prediction model, and the model outputs the prediction result of pipe burst risk according to the input data;

[0042] The prediction result includes outputting a probability value indicating whether there is a pipe burst risk or a classification label indicating whether there is a pipe burst risk. By setting the recognition threshold, it is determined whether there is a pipe burst risk.

[0043] S4. If it is determined that there is a risk of pipe burst, the alarm mechanism will be triggered and a warning message will be sent to the supervisor. At the same time, according to the location and severity of the pipe burst, the corresponding emergency response process can be initiated as appropriate, including closing the isolation valve and starting the emergency drainage system to prevent the pipe burst from expanding and reduce losses.

[0044] Furthermore, the pipe burst risk prediction model includes continuous optimization based on historical data and abnormal results from manual feedback to improve the accuracy and efficiency of early warning.

[0045] The method includes establishing a mutual verification mechanism for pressure data collected by a high-frequency pressure meter, sound data collected by a hydrophone, and flow data collected by a flow meter, so as to overcome data anomalies caused by equipment failure.

[0046] The above-mentioned mutual verification mechanism includes localized deployment. First, an abnormal data group [s, a, b, c] is constructed according to the position relationship of the collection devices in the data collection unit, where s represents the position, which contains three types of collection devices, and a, b, and c represent the values ​​collected by the three types of collection devices respectively. Then, it is judged on the local node whether a, b, and c are within the normal range. If so, the abnormal data group is sent to the cloud server side for processing by the pipe burst risk prediction model, thereby reducing the consumption of computing resources on the cloud server side.

[0047] Beneficial effects: Through real-time monitoring and intelligent analysis, the present invention can effectively identify and warn of pipe burst risks in underground integrated pipe corridors, significantly improve the safety and reliability of pipe corridor operations, and reduce economic losses and social impacts caused by pipe burst accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a system architecture diagram described in the present invention. DETAILED DESCRIPTION

[0049] The implementation of the present invention includes installing equipment and integrating the system in an existing pipe gallery, or directly using the pipe burst warning system of the present invention when building a new pipe gallery. During specific implementation, appropriate adjustments and optimizations can be made according to the actual situation and specific needs of the pipe gallery.

[0050] The underground comprehensive pipe gallery pipe burst early warning system provided by the present invention mainly consists of four parts: a data acquisition unit, a data transmission unit, a data analysis unit and an early warning response unit.

[0051] (1) Data acquisition unit

[0052] High-frequency pressure gauges: High-frequency pressure gauges are installed at key locations of the water supply pipeline (such as bends, branch points, vulnerable sections, etc.) at intervals of 0.8-1.0km to monitor and identify water hammer (pressure transient) signals, including water pressure fluctuations caused by valve opening, water pump shutdown, and flow fluctuations caused by large water users, and transmit the monitoring data to the data analysis unit.

[0053] Hydrophone: Also installed at the above-mentioned key positions, it is used in conjunction with a high-frequency pressure gauge to collect sound data in the pipe network to identify whether there is a pipe burst or abnormal noise (such as water leakage, vibration, etc.).

[0054] Flowmeter: Installed on the straight section of the pipeline at intervals of 0.8 to 1.0 km, avoid installation on elbows, valves and other areas where eddies or turbulence may occur. This ensures the accuracy of flowmeter measurement and avoids measurement errors caused by poor fluid flow conditions. The main function is to monitor the fluid flow in the pipeline in real time to provide more comprehensive information on the pipeline operation status.

[0055] (2) Data transmission unit

[0056] The pressure data and sound data collected by the data acquisition unit are transmitted to the data analysis unit in real time by wired or wireless transmission. To ensure the stability and reliability of data transmission, the system can be equipped with a backup transmission channel to prevent the main channel from failing.

[0057] (3) Data analysis unit

[0058] The unit receives data from the data transmission unit, and processes the pressure, sound and flow data through a signal analysis algorithm to extract characteristic parameters.

[0059] The extracted characteristic parameters are analyzed using a trained machine learning algorithm to identify pressure transients, abnormal noise, and flow changes in the pipeline, thereby determining whether there is a risk of pipe burst.

[0060] The data analysis unit also has self-learning capabilities and can continuously optimize algorithms based on historical data to improve the accuracy and efficiency of early warnings.

[0061] For the pressure data collected by the high-frequency pressure meter, the characteristics of the pressure change, such as the amplitude, frequency, and duration, can be extracted; for the sound data collected by the hydrophone, the spectral characteristics and energy distribution of the sound signal can be extracted. Combining the data of the high-frequency pressure meter and the hydrophone, the flow information provided by the flow meter can be mutually verified, further enriching the decision-making basis for pipe burst warning.

[0062] (4) Early warning response unit

[0063] Once the data analysis unit determines that there is a risk of pipe burst, the early warning response unit will be activated immediately to send early warning information to relevant personnel. Early warning information can be sent through SMS, email, APP push and other methods to ensure that relevant personnel can receive it in time.

[0064] The early warning response unit also has a remote control function, which can automatically close relevant valves and cut off the water source when necessary to prevent the pipe burst from expanding.

[0065] Further, in combination with the implementation method, the implementation steps of the present invention are as follows:

[0066] Step 1: Data Collection

[0067] The high-frequency pressure gauge, hydrophone and flow meter collect the pressure, sound and flow data of the pipeline in real time, and send the data to the data transmission unit regularly (such as every minute).

[0068] Step 2: Data Transmission

[0069] The data transmission unit transmits the collected data to the data analysis unit via wired or wireless means.

[0070] Step 3: Data processing

[0071] The collected data is preprocessed through the signal analysis algorithm to extract characteristic parameters, which are then input into the machine learning model to obtain the pipe burst prediction model.

[0072] For the pressure data collected by the high-frequency pressure meter, the characteristics of the pressure change, such as the amplitude, frequency, and duration, can be extracted; for the sound data collected by the hydrophone, the spectral characteristics and energy distribution of the sound signal can be extracted. Combining the data of the high-frequency pressure meter and the hydrophone, the flow information provided by the flow meter can be mutually verified, further enriching the decision-making basis for pipe burst warning.

[0073] The preprocessing steps include data cleaning (removing outliers, noise, etc.), data compression (reducing the amount of data and improving processing efficiency), and data conversion (such as converting analog signals into digital signals).

[0074] Step 4: Risk Identification

[0075] Statistical analysis, signal processing technology and other methods are used to analyze the extracted characteristic parameters and identify possible abnormal patterns. For example, by comparing the difference between the current pressure and flow changes and the historical data, it can be determined whether there are abnormal pressure and flow fluctuations; through the spectrum analysis of the sound signal, abnormal water flow or leakage sound can be identified.

[0076] Based on a large amount of historical data (including data in normal and burst states), a machine learning model is trained to learn the complex relationships in the data and build a prediction model for the risk of burst.

[0077] The real-time collected and preprocessed data is input into the trained machine learning model, and the model outputs the prediction result of pipe burst risk based on the input data. The prediction result may be a probability value (indicating the high or low risk of pipe burst) or a classification label (indicating whether there is a risk of pipe burst). By setting the recognition threshold, the system determines whether there is a risk of pipe burst.

[0078] Step 5: Early warning response

[0079] If there is a risk of pipe burst, the alarm mechanism will be triggered and warning information will be sent to relevant personnel. Warning information can be sent through various channels, such as SMS, email, APP push, etc., to ensure that relevant personnel can receive it in time.

[0080] At the same time, according to the location and severity of the pipe burst, the corresponding emergency response process can be initiated as appropriate, such as closing the isolation valve, starting the emergency drainage system, etc., to prevent the pipe burst from expanding and reduce losses.

[0081] Step 6: On-site disposal

[0082] Relevant personnel who receive the warning information should immediately go to the scene to verify the situation and take appropriate measures to deal with it. If it is confirmed to be a pipe burst, emergency repair work should be organized quickly to restore the normal operation of the water supply pipeline.

[0083] Step 7: Continuous Optimization and Improvement

[0084] Feedback mechanism: Establish a feedback mechanism for the early warning system to collect actual conditions and effect feedback of early warning responses to evaluate the accuracy and effectiveness of the early warning system.

[0085] Model update: Based on feedback results and new data samples, the machine learning model is continuously updated and optimized to improve the prediction accuracy and robustness of the early warning system.

[0086] In the above method implementation and model construction, specifically:

[0087] (I) Signal data analysis

[0088] Step 1. Data preprocessing

[0089] Data cleaning: Remove outliers, noise, etc. to improve data quality. This includes removing abnormal data points due to equipment failure or environmental factors.

[0090] Data compression: Reduce the amount of data and improve processing efficiency. Through data compression algorithms, the storage and transmission costs of data can be reduced while retaining key information.

[0091] Data conversion: such as converting analog signals into digital signals to facilitate subsequent digital signal processing and analysis.

[0092] Step 2: Feature extraction

[0093] 1) Pressure data

[0094] Pressure change magnitude: Calculate the difference between adjacent data points to obtain the magnitude of the pressure change:

[0095] DeltaP=P(t i+1 )-P(t i )

[0096] Among them, P(t i ) represents the time point t i The pressure value;

[0097] Pressure change frequency: The frequency distribution of pressure change is obtained by performing Fourier transform or fast Fourier transform on the pressure data:

[0098]

[0099] Among them, P(f) is the component of pressure at frequency f, and P(t) is the pressure signal in the time domain.

[0100] Duration of pressure change: record the length of time when the pressure exceeds a certain threshold, including extracting characteristic parameters such as the amplitude, frequency, and duration of pressure change; this characteristic parameter can reflect the change of pressure in the pipeline and help identify water hammer (pressure transient) signals;

[0101] 2) Sound data characteristics

[0102] Spectrum analysis: By performing Fourier transform on the sound signal, the spectrum characteristics of the sound are obtained, so as to identify abnormal sounds including water flow and leakage.

[0103]

[0104] Among them, S(f) is the component of the sound at frequency f, and s(t) is the sound signal in the time domain;

[0105] Energy distribution: Calculate the energy distribution of the sound signal at each frequency to identify the sound characteristics:

[0106] E(f)=|S(f)| 2

[0107] Where E(f) is the sound energy at frequency f;

[0108] Extract the spectrum characteristics and energy distribution of the sound signal, and through spectrum analysis, identify abnormal water flow or leakage sounds, which are usually associated with pipe burst events;

[0109] 3) Traffic data characteristics

[0110] Flow rate change amplitude: Similar to the pressure change amplitude, calculate the difference between adjacent data points:

[0111] DeltaQ=Q(t i+1 )-Q(t i )

[0112] Among them, Q(t i ) represents the time point t i The flow value;

[0113] Flow rate of change: Calculate the derivative of the flow to obtain the flow rate change trend:

[0114]

[0115] By extracting characteristic parameters including the amplitude and frequency of flow changes, flow data can reflect the flow state of the fluid in the pipeline and help determine whether there are abnormal flow fluctuations.

[0116] 2. Machine Learning Process

[0117] The machine learning algorithm is used to analyze the extracted feature parameters and predict the risk of burst in water supply pipelines.

[0118] Model training: Train a machine learning model based on a large amount of historical data (including data in normal and burst states). These historical data should cover various pipeline operating conditions and burst event types.

[0119] Machine learning models can choose algorithms such as support vector machines, neural networks, random forests, or gradient boosting trees. These algorithms are able to learn complex relationships in the data and build a predictive model for pipe burst risk.

[0120] Model prediction: The real-time collected and pre-processed data is input into the trained machine learning model, and the model outputs the prediction results of pipe burst risk based on the input data.

[0121] The prediction result may be a probability value (indicating the risk of pipe burst) or a classification label (indicating whether there is a risk of pipe burst). By setting the recognition threshold, the system can determine whether there is a risk of pipe burst.

[0122] Model optimization: The machine learning model has self-learning capabilities and can continuously optimize the algorithm based on historical data and feedback results to improve the accuracy and efficiency of early warning.

[0123] By establishing a feedback mechanism for the early warning system, the actual situation and effect feedback of the early warning response are collected to evaluate the accuracy and effectiveness of the early warning system.

[0124] Based on feedback results and new data samples, the machine learning model is continuously updated and optimized to adapt to changes in pipeline operating conditions and new developments in pipe burst event types.

[0125] Finally, the implementation of the present invention can achieve the following significant effects:

[0126] Early warning: Through real-time monitoring and data analysis, early warning of pipe burst incidents can be achieved to reduce the impact of emergencies on pipeline corridors and urban infrastructure.

[0127] Accurate positioning and early warning: Using data from high-frequency pressure gauges, hydrophones and flow meters, the location of pipe bursts can be accurately determined, improving the efficiency of emergency repairs. Advanced algorithms are used to analyze the collected data, accurately determine the risk of pipe bursts, and reduce the occurrence of false alarms and missed alarms.

[0128] Efficient response: The early warning response unit can quickly initiate the early warning process and send early warning information to relevant personnel in a variety of ways to ensure timely response and disposal.

[0129] Intelligent management: Combined with machine learning algorithms, it can realize intelligent monitoring and management of pipeline status and improve management level.

[0130] Remote control: With remote control function, it can automatically close relevant valves when necessary, cut off the water source and prevent the expansion of pipe burst incidents.

Claims

1. An underground comprehensive pipe gallery pipe burst warning system, characterized in that: The system includes a data acquisition unit, a data transmission unit, a data analysis unit, and an early warning response unit. It can monitor and analyze key parameters of water supply pipelines in real time, accurately identify potential pipe burst risks through signal processing and machine learning algorithms, and issue early warning signals in a timely manner. Data acquisition unit: including high-frequency pressure gauge, hydrophone and flow meter, real-time monitoring of pipeline pressure, sound and flow data; Data transmission unit: used to transmit the data collected by the data collection unit to the data analysis unit, and the data transmission unit includes wired connection or wireless transmission; Data analysis unit: Receives data from the data acquisition unit through the data transmission unit, and processes the pressure, sound and flow data through the signal analysis algorithm, including obtaining characteristic parameters; also includes using the trained machine learning model to analyze the characteristic parameters to identify pressure transients, abnormal noise and flow changes in the pipeline, so as to determine whether there is a risk of pipe burst; Early warning response unit: If the data analysis unit determines that there is a risk of pipe explosion, the early warning response unit will be activated immediately and send early warning information to the supervisor.

2. The underground comprehensive pipe gallery pipe burst warning system according to claim 1 is characterized in that: The data analysis unit includes the following data processing: (1) Pressure data Pressure change magnitude: Calculate the difference between adjacent data points to obtain the magnitude of the pressure change: DeltaP=P(t i+1 )-P(t i ) Among them, P(t i ) represents the time point t i The pressure value; Pressure change frequency: The frequency distribution of pressure change is obtained by performing Fourier transform or fast Fourier transform on the pressure data: Among them, P(f) is the component of pressure at frequency f, and P(t) is the pressure signal in the time domain; Duration of pressure change: record the length of time when the pressure exceeds a certain threshold, including extracting characteristic parameters such as the amplitude, frequency, and duration of pressure change; this characteristic parameter can reflect the change of pressure in the pipeline and help identify water hammer signals; (2) Sound Data Characteristics Spectrum analysis: By performing Fourier transform on the sound signal, the spectrum characteristics of the sound are obtained, so as to identify abnormal sounds including water flow and leakage. Among them, S(f) is the component of the sound at frequency f, and s(t) is the sound signal in the time domain; Energy distribution: Calculate the energy distribution of the sound signal at each frequency to identify the sound characteristics: E(f)=∣S(f)∣ 2 Where E(f) is the sound energy at frequency f; Extract the spectrum characteristics and energy distribution of the sound signal, and through spectrum analysis, identify abnormal water flow or leakage sounds, which are usually associated with pipe burst events; (3) Traffic data characteristics Flow rate change amplitude: Similar to the pressure change amplitude, calculate the difference between adjacent data points: DeltaQ=Q(t i+1 )-Q(t i ) Among them, Q(t i ) represents the time point t i The flow value; Flow rate of change: Calculate the derivative of the flow to obtain the flow rate change trend: By extracting characteristic parameters including the amplitude and frequency of flow changes, flow data can reflect the flow state of the fluid in the pipeline and help determine whether there are abnormal flow fluctuations.

3. The underground comprehensive pipe gallery pipe burst warning system according to claim 1 is characterized in that: The data acquisition unit includes a high-frequency pressure gauge, a hydrophone, and a flow meter; The high-frequency pressure gauge and hydrophone are arranged at the turning point, branch point, vulnerable section, and the same point. The high-frequency pressure gauge collects the flow fluctuation caused by the opening of the valve, the start and stop of the water pump, and the water use of the user, as well as the water pressure fluctuation, and transmits the monitoring data to the data analysis unit; the hydrophone is used in conjunction with the high-frequency pressure gauge to identify whether there is a pipe burst or abnormal noise by collecting sound data in the pipe network. The abnormal noise includes water leakage sound and vibration data; The flow meter is arranged at a non-turning or valve position to avoid areas of eddy current or turbulence in the corresponding pipe gallery.

4. The underground comprehensive pipe gallery pipe burst warning system according to claim 1 is characterized in that: The early warning response unit includes a remote control module to realize control of valves, and the pipe gallery includes one or more control valves; The early warning response unit also includes sending early warning information to supervisors, and the sending of early warning information includes SMS, email, and APP push.

5. An early warning method for the underground utility corridor pipe burst early warning system as claimed in claim 1, characterized in that: The following steps are involved: S1. Deploy data acquisition units on the water supply pipelines in the integrated pipe gallery, including high-frequency pressure gauges, hydrophones and flow meters, to monitor pipeline pressure, sound and flow data in real time, and pre-process the collected data, including cleaning and denoising, data compression and data conversion, so as to reduce the cost of data transmission and storage, and obtain data that can be used for signal analysis algorithm recognition; S2. Use a combination of machine learning models and signal analysis algorithms to build a pipe burst risk prediction model. The construction of the model includes using statistical analysis and signal processing technology to perform abnormal analysis on the data collected from the water supply pipe, including feature extraction of the data collected by the data acquisition unit, and inputting the processed data into the trained machine learning model to obtain a pipe burst risk prediction model to identify possible abnormal patterns, including comparing the current pressure and flow changes with the historical data to determine whether there are abnormal pressure and flow fluctuations; and identifying abnormal water flow sounds or leakage sounds through spectrum analysis of sound signals; S3, inputting the real-time collected and pre-processed data into the trained pipe burst risk prediction model, and the model outputs the prediction result of pipe burst risk according to the input data; The prediction result includes outputting a probability value indicating whether there is a pipe burst risk or a classification label indicating whether there is a pipe burst risk. By setting the recognition threshold, it is determined whether there is a pipe burst risk. S4. If it is determined that there is a risk of pipe burst, the alarm mechanism will be triggered and a warning message will be sent to the supervisor. At the same time, according to the location and severity of the pipe burst, the corresponding emergency response process can be initiated as appropriate, including closing the isolation valve and starting the emergency drainage system to prevent the pipe burst from expanding and reduce losses.

6. The early warning method of the underground utility corridor pipe burst early warning system according to claim 5 is characterized in that: The pipe burst risk prediction model includes continuous optimization based on historical data and abnormal results of manual feedback to improve the accuracy and efficiency of early warning.

7. The early warning method of the underground utility corridor pipe burst early warning system according to claim 5 is characterized in that: The method includes establishing a mutual verification mechanism for pressure data collected by a high-frequency pressure meter, sound data collected by a hydrophone, and flow data collected by a flow meter, so as to overcome data anomalies caused by equipment failure.

8. The early warning method of the underground utility gallery pipe burst early warning system according to claim 7 is characterized in that: The mutual verification mechanism includes localized deployment. First, an abnormal data group [s, a, b, c] is constructed according to the position relationship of the collection devices in the data collection unit. s represents the position, which contains three types of collection devices. A, b, and c represent the values ​​collected by the three collection devices respectively. Then, it is judged on the local node whether a, b, and c are within the normal range. If so, the abnormal data group is sent to the cloud server and processed by the pipe burst risk prediction model, thereby reducing the consumption of computing resources on the cloud server.

Images