Artificial intelligence leakage distinguishing and positioning method for municipal water supply and drainage pipe network

Abstract

The invention relates to the technical field of pipeline water leakage monitoring, and discloses an artificial intelligence leakage distinguishing and positioning method for a municipal water supply and drainage pipe network, and the method comprises the following steps: 1, data collection; 2, analyzing a water leakage detection data model; 3, establishing a leakage point positioning model; step 4, alarm and work order generation; and 5, on-site confirmation and maintenance are carried out. By introducing artificial intelligence and big data technologies, all-day real-time monitoring of a pipe network is realized, the efficiency and accuracy of leakage detection are improved, a leakage point positioning algorithm model is established, accurate positioning of a leakage point is realized, the maintenance time and cost are reduced, and meanwhile, the leakage rate of the pipe network is reduced through real-time monitoring and timely maintenance, so that the maintenance efficiency of the pipe network is improved. The production-marketing ratio and the water-saving rate of a water company are improved, an algorithm model is continuously optimized, and the leakage identification and positioning precision is improved, so that the pipe network monitoring requirements in different scenes and complex geological environments are met.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline leakage monitoring, and specifically to a method for artificial intelligence leakage identification and location of municipal water supply and drainage pipe networks. Background Technique

[0002] In the management and maintenance of municipal water supply and drainage pipe networks, the leakage problem has always been a difficulty and key point in the industry. Traditional leakage detection methods, such as manual inspections and pressure monitoring, have problems such as low detection efficiency and inaccurate location, and it is difficult to meet the requirements of modern cities for efficient and precise management of water supply pipe networks. With the rapid development of technologies such as the Internet of Things, big data, and artificial intelligence, intelligent leakage identification and location technologies have emerged, bringing a revolutionary change to the management of municipal water supply and drainage pipe networks.

[0003] Existing intelligent leakage identification and location technologies mainly fall into two categories based on acoustic principles and non-acoustic principles. The acoustic principle mainly involves installing acoustic sensors on the pipeline to collect the audio data of the water flow in the pipeline, and then using big data algorithm models to analyze and process the audio data to identify the leakage signal and determine the location of the leakage point. This method has advantages such as high detection efficiency and accurate location, and has become one of the current mainstream technologies.

[0004] The non-acoustic principle mainly relies on changes in hydraulic parameters such as pressure and flow to detect leakage. For example, through transient analysis or pressure monitoring and other methods, abnormal changes in the water pressure or flow in the pipeline can be detected, thereby inferring whether there is a leakage situation. However, this method usually can only determine the approximate area of the leakage and cannot accurately locate the leakage point, so there are certain limitations in practical applications. Therefore, we propose a method for artificial intelligence leakage identification and location of municipal water supply and drainage pipe networks. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for artificial intelligence leakage identification and location of municipal water supply and drainage pipe networks. By introducing artificial intelligence and big data technologies, real-time all-day monitoring of the pipe network is achieved, the efficiency and accuracy of leakage detection are improved, a leakage point location algorithm model is established to achieve precise location of the leakage point, the repair time and cost are reduced. At the same time, through real-time monitoring and timely repair, the leakage rate of the pipe network is reduced, the production and sales ratio and water saving rate of the water service company are improved, and the algorithm model is continuously optimized to improve the accuracy of leakage identification and location to meet the pipe network monitoring requirements in different scenarios and complex geological environments, solving the problems raised in the background technique.

[0006] To achieve the above purpose, the present invention provides the following technical solution: A method for artificial intelligence leakage identification and location of municipal water supply and drainage pipe networks, including the following steps:

[0007] Step 1: Data collection: High-sensitivity sensors installed on pipelines or valves are used to collect real-time audio data of the water flow in the pipelines, and the data is transmitted to the data processing center through Internet of Things technology;

[0008] Step 2: Leak detection data model analysis: Feature extraction and selection are performed on the collected audio data, and after on-site confirmation, it is marked to form a data set containing leakage data and non-leakage data;

[0009] Step 3: Leak point location model establishment: Extract the features of the audio data, use a machine learning classifier for model training and cross-validation, select the best hyperparameters to establish a leak detection data model, design a custom function to process the audio data, through frequency domain feature analysis, anomaly processing and smoothing processing, combined with the pipeline length to calculate the distance from the leak point to the sensor, and establish a leak point location model;

[0010] Step 4: Alarm and work order generation: Once a leak signal is detected, the system automatically generates an alarm message and uploads it to the work order business system of the water service company;

[0011] Step 5: On-site confirmation and repair: The leak detection personnel and repair personnel of the water service company arrive at the site according to the positioning information, reconfirm the leak situation, excavate near the positioning point, and find and repair the actual leak point.

[0012] As a preferred embodiment of the present invention, the sensitivity of the high-sensitivity sensor in Step 1 is not less than 1200 pC / g, and the sampling rate is 8192 Hz.

[0013] As a preferred embodiment of the present invention, the data collection time in Step 1 is selected during the period with the least environmental interference.

[0014] As a preferred embodiment of the present invention, the features extracted in Step 3 include linear time domain features, frequency domain features, and features describing the non-linear features of the leakage signal.

[0015] As a preferred embodiment of the present invention, the machine learning classifier used in Step 3 is selected from at least one of support vector machine (SVM), decision tree (DT), K-nearest neighbor (KNN), extreme gradient boosting (XGBoost), and light gradient boosting (LightGBM).

[0016] As a preferred embodiment of the present invention, the custom function in Step 3 includes short-time Fourier transform (STFT), smoothing processing, root mean square (RMS) calculation, and maximum correlation analysis (MRCA).

[0017] As a preferred embodiment of the present invention, the alarm information in step four includes information such as the detection device number, installation address, water leakage identification, and water leakage location.

[0018] As a preferred embodiment of the present invention, after the repair is completed in step five, the repair result is returned to the intelligent water system to form a closed-loop management.

[0019] As a preferred embodiment of the present invention, it also includes data analysis and optimization: based on data samples, deep learning methods are used to autonomously detect leakage signals to improve the accuracy of identification.

[0020] As a preferred embodiment of the present invention, it also includes step system security and maintenance: ensuring the security of data transmission, storage, and processing, and regularly maintaining and upgrading the system to ensure the stable operation of the system and the accuracy of data.

[0021] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0022] By introducing artificial intelligence and big data technologies, the present invention realizes all-day real-time monitoring of the pipe network, improves the efficiency and accuracy of leakage detection, establishes a leakage point positioning algorithm model, realizes precise positioning of the leakage point, reduces repair time and costs, and at the same time, by real-time monitoring and timely repair, reduces the leakage rate of the pipe network, improves the production and sales ratio and water conservation rate of the water company, and continuously optimizes the algorithm model to improve the accuracy of leakage identification and positioning to meet the pipe network monitoring requirements in different scenarios and complex geological environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] By reading the detailed description of the non-limiting embodiments with reference to the following drawings, other features, objects, and advantages of the present invention will become more apparent:

[0024] Figure 1 It is a flowchart of a method for artificial intelligence leakage identification and positioning of a municipal water supply and drainage pipe network of the present invention.

[0025] Figure 2 It is a schematic diagram of the feature importance of the XGBoost model for artificial intelligence leakage identification and positioning of a municipal water supply and drainage pipe network of the present invention.

[0026] Figure 3 It is a schematic diagram of the prediction shared by a single sample explained by the SHAP value of the XGBoost model for artificial intelligence leakage identification and positioning of a municipal water supply and drainage pipe network of the present invention.

[0027] Figure 4 It is a schematic diagram of the PDP value of the XGBoost model for artificial intelligence leakage identification and positioning of a municipal water supply and drainage pipe network of the present invention.

[0028] Figure 5 Schematic diagram of the decision tree of the XGBoost model for artificial intelligence leakage identification and location of a municipal water supply and drainage pipe network according to the present invention.

[0029] Figure 6 Schematic diagram of the LIME local interpretation method of the XGBoost model for artificial intelligence leakage identification and location of a municipal water supply and drainage pipe network according to the present invention. Specific implementation manners

[0030] In order to make the technical means, creative features, achieved purposes and functions of the present invention easy to understand, the present invention will be further described below in conjunction with specific implementation manners.

[0031] The municipal water supply pipeline leakage monitoring and location system includes four major parts: a management terminal (PC), a mini-program terminal, a big data algorithm model terminal, and hardware devices. Combining a software platform, a big data algorithm model, and Internet of Things devices, it realizes the functions of real-time leakage monitoring and accurate location of pipe networks in various scenarios such as municipal pipe networks, community pipe networks, and in-building pipes.

[0032] The overall architecture of the system is as follows:

[0033] Infrastructure system, fully applying the infrastructure of the data platform to provide unified, stable, and efficient infrastructure resources for this system, mainly providing capabilities such as network, storage, and computing.

[0034] Application support system, composed of a GIS engine, an object storage service, a notification push service, a short message service, etc. Build data models and business models according to the needs of scenario applications and business collaboration to achieve interconnection of relevant business systems, data sharing, and business collaboration.

[0035] Policy and system system, improve various supporting management systems, revise regulatory documents that do not match digital reform, and optimize the construction environment of government digital reform.

[0036] Organizational guarantee system, strengthen organizational guarantee, establish leadership responsibility systems for intelligent distribution direct sharing planning, construction, operation and maintenance, and implement the method of "project implementation + special class promotion", improve the efficient coordination mechanism across departments, fields, and levels, and enhance the digital capabilities of reform entities.

[0037] Network security system, coordinate development and security, establish the bottom-line thinking of network security, strictly implement the requirements of hierarchical protection, improve the security protection system for critical information infrastructure, the security protection system for public data and personal information, improve the network security technical protection system covering physical facilities, networks, platforms, applications, and data, enhance the active defense ability, monitoring and early warning ability, emergency response ability, and collaborative governance ability of network security, and build a network security barrier.

[0038] The data collected by the Internet of Things devices deployed at various points are preliminarily parsed, cleaned, and analyzed by an algorithm model at the aggregation point, and then synchronized to the private network of the water company through the boundary platform.

[0039] On the premise of ensuring data and network security, aggregation is carried out through the private network of the water company according to the actual network operating environment of different subsystems. For the information systems built in the private network of the water company, aggregation, cleaning, structuring, and algorithm analysis are completed in the private network of the water company, and data is synchronized from the private network of the water company through the boundary platform.

[0040] System Technical Principle

[0041] 1. Technical Principle of the Management End and Mini Program Development

[0042] In the system, the development technical principles of the management end and the mini program end are as follows:

[0043] Development Mode: Agile Development;

[0044] Development Method: Front-end and Back-end Separation Development;

[0045] Operating System: Domestic Kirin V10;

[0046] Development Languages and Technical Frameworks: JAVA, Spring Boot, jdk1.8, Vue, uniapp (mini program), H5 (mini program), facilitating fast development and version iteration and upgrade;

[0047] Application Middleware: Redis, Nginx, RabbitMQ;

[0048] Application Database: Domestic DM Database (DM);

[0049] 2. Technical Principle of Big Data Algorithm Model

[0050] The big data algorithm model is divided into a leakage monitoring algorithm model and a leakage point location algorithm model. Combining the analysis results of the two algorithm models, real-time monitoring and accurate positioning of pipeline leakage are achieved.

[0051] This project studies the accuracy of leakage identification through two sets of analysis methods. One set is based on the extracted signal features and uses a classifier for classification, and the other set is based on data samples (such as time series and time-frequency images) and uses deep learning methods to autonomously detect leakage signals. Generally speaking, features based on linear time domain and frequency domain are extracted, and the optimal feature subset for classification is determined by performing short-time stationary processing on the leakage signal, but little attention is paid to the non-linear features used to describe the non-stationary features contained in the leakage signal.

[0052] The company uses sensor data for leak detection in a real water pipe network scenario, and builds an application of a traditional machine learning model to identify leaks using vibration signals collected by wireless piezoelectric accelerometers in real and complex pipe network scenarios, achieving three goals:

[0053] Construct a signal dataset using the data collected by accelerometers in the actual water pipe network;

[0054] Extract features that describe the non-linear characteristics of the leak signal as the input of the classifier to discover discriminant information;

[0055] Test and compare the leak detection performance of machine learning models.

[0056] The present invention provides a technical solution: a method for artificial intelligence leak identification and location in a municipal water supply and drainage pipe network. High-sensitivity sensors are installed on pipes or valves to collect real-time audio data of the water flow in the pipes. The sensitivity of the sensors used is not less than 1200 pC / g, and the sampling rate is 8192 Hz. The piezoelectric accelerometer is connected to the pipe or valve through a magnetic seat in the water distribution network. Different from the related technologies, the sensors are independent of each other, and their measurement results are transmitted to the data processing center through the Internet of Things (IOT). The data acquisition time can be set manually, and the time period with the least environmental interference is selected as 2:00 - 4:00 h.

[0057] Generally, once the data acquisition platform identifies a suspected leak signal from the data transmitted by the sensors, the staff will go to the site to confirm whether the pipe is leaking. After excavating and determining the leak and possible repair situation of the pipe, the sensor data of the suspected leak is marked as leak data. When there is no suspected leak data on the platform, the data collected by the sensors during the normal operation of the pipe network is collected and marked as non-leak data.

[0058] Traditional machine learning classification algorithms are mainly used for modeling and classification of structured data. Structured data generally uses rows as sample numbers and columns as feature names. Therefore, the R & D department has collected a large number of audio data samples, and then extracts multiple feature values in the time-frequency domain to train a model of a common classifier. The following are the methods for each feature value:

[0059] 1. Root Mean Square (RMS): mainly used to reflect the smoothness and energy magnitude of the signal

[0060]

[0061] 2. Zero Crossing Rate (ZCR): mainly used to reflect the smoothness and severity of signal changes

[0062]

[0063] 3. Teager Energy Operator (TEO): mainly used to extract the energy features of signals

[0064]

[0065] 4. Normalized autocorrelation kurtosis (Kur): mainly used to judge the non - linear and periodic characteristics of signals

[0066]

[0067] 5. Approximate entropy (ApEn): mainly used to evaluate the regularity, noise sensitivity and non - linear characteristics of signals

[0068] ApEn(e, r, N) = C e (r) - C e+1 (r)

[0069] 6. Sample entropy (SampEn): an improvement of approximate entropy, more suitable for evaluating the complexity and uncertainty of signals

[0070]

[0071] 7. Additionally, it also includes other time - domain features such as other fractal dimensions (FD), Hurst exponent, and frequency - domain features such as bispectral entropy (BspEn), higher - order spectrum (HOS).

[0072] The above eigenvalues are used to calculate a classification dataset for each piece of valid data through a designed algorithm, and then the most suitable several features are found through one - way ANOVA.

[0073] Model building

[0074] After feature engineering, a classification dataset is obtained. Since the dimensions of different eigenvalues are different, a normalization operation is first performed to obtain a classification dataset where all eigenvalues are between 0 and 1. Then the dataset is randomly split into a training set, a validation set, and a test set in a ratio of 8:1:1. Model training and cross - validation are performed using the following traditional machine learning algorithms: Support Vector Machine (SVM), Decision Tree (DT), k - Nearest Neighbor (KNN), Ensemble Learning (xgboost, lightgbm).

[0075] Finally, the model is evaluated based on five metrics after testing. By comparison, it is found that xgboost has the best effect. Therefore, the model trained by xgboost is finally selected for leak detection. The five evaluation metrics are Accuracy, Specificity, Sensitivity, Precision, F1score, and their calculation formulas are as follows:

[0076]

[0077] Where TP / TN / FN / FP are the numbers of true positives, true negatives, false negatives, and false positives, respectively.

[0078] XGBoost is an extension based on Gradient Boosting Decision Trees (GBDT), but it significantly improves the efficiency and performance of the algorithm through various optimization means.

[0079] Regularization

[0080] L1 and L2 regularization: XGBoost supports L1 (Lasso) and L2 (Ridge) regularization, which can prevent the model from overfitting.

[0081] Tree complexity control: By restricting parameters such as the depth of the tree and the number of leaf nodes, XGBoost can help control the complexity of the model, thereby improving the generalization ability of the model.

[0082] Parallel processing

[0083] XGBoost supports parallel processing and can run efficiently on multi-core processors. It greatly improves the training speed by parallelizing the search for split points of features.

[0084] Handling missing values

[0085] XGBoost has a built-in mechanism for handling missing values. The algorithm can automatically learn how to handle missing values, thus reducing the dependence on data preprocessing.

[0086] Flexible objective function

[0087] XGBoost allows users to customize the objective function, which makes it applicable not only to regression and classification tasks but also to other types of tasks such as ranking and probability estimation.

[0088] Built-in cross-validation

[0089] XGBoost supports cross-validation, and users can conveniently evaluate the model performance and tune the parameters.

[0090] High performance

[0091] XGBoost is optimized in terms of memory usage and computing speed, making it perform excellently on large-scale datasets. It uses some advanced algorithms and data structures such as the Approximate Algorithm and Histogram-based split point search to accelerate the calculation.

[0092] Interpretability

[0093] Although XGBoost is a black-box model, it provides some tools (such as feature importance) to help users understand how the model works. Specifically as follows:

[0094] (1) Feature importance.

[0095] (2) SHAP values explain the predictions shared by individual samples.

[0096] (3) PDP values (Partial Dependence Plot visualization tool for showing how a single feature affects the model's prediction results).

[0097] (4) Decision tree visualization, visualizing the structure of individual decision trees to understand how the model makes decisions.

[0098] (5) LIME local interpretation method, explaining predictions by generating a simplified model around individual samples.

[0099] The advanced nature of XGBoost is reflected in its efficient algorithm, flexible parameter settings, high-performance computing power, and wide range of application scenarios. It has become one of the preferred tools for data scientists and machine learning engineers, especially in scenarios that require high performance and high flexibility.

[0100] Establish a leak point location model

[0101] 1. Design custom functions

[0102] Design custom functions for various different data processing to convert audio data into energy index values.

[0103] (1) Short-time Fourier transform custom function (stft):

[0104] Set the frame length, window length, calculate the number of audio frames, and then perform a fast Fourier transform (FFT) through frame segmentation and windowing to obtain a complex result and take the modulus to get the amplitude of each frame signal, that is, the signal intensity, and finally return a two-dimensional matrix corresponding to the number of frames and signal intensity.

[0105] (2) Smoothing custom function (smoothing)

[0106] Set the effective smoothing factor, filter out high-frequency jumping noise, and overwrite the original values.

[0107] (3) Root mean square custom function (rms)

[0108] Sum the squares of each value in the sequence, divide by the number of sequence items, and then take the square root.

[0109] (4) mrca custom function (mrca)

[0110] By dividing the audio data into multiple windows and performing simple local anomaly processing within each window. The logic of the anomaly processing is based on whether the value of the current point is less than a certain multiple (determined by factor) of the local average. If it exceeds this multiple, the current point is considered noise and is replaced with the value of the previous point. Finally, the processed audio data is returned.

[0111] 2. Calculate the distance to the leakage point

[0112] Based on the above function capabilities, it can be used to calculate the distance from the leakage point to the device. The specific process is as follows:

[0113] (1) Obtain the paired audio signals of the telecommunication base station for synchronous timekeeping.

[0114] (2) Use the stft function to convert the audio data left_wav and right_wav collected by the left and right microphones into a short-time Fourier transform matrix, and calculate the number of frequency components of the obtained STFT matrix.

[0115] (3) Create and initialize arrays left_stft, right_stft, left_stft_smooth, and right_stft_smooth with the same size as the STFT matrix, which are used to store the frame energies after local processing and after smoothing processing.

[0116] (4) Create and initialize the left_rms and right_rms lists, which are used to store the rms values of each frequency component.

[0117] (5) Create and initialize left_rms_value and right_rms_value, which are used to store the total rms value and for subsequent calculation of the average rms value.

[0118] (6) Apply the mrca function to each frequency component for local anomaly processing, and then apply the smoothing function for smoothing processing.

[0119] (7) Calculate the rms value of the smoothed signal and accumulate it into left_rms_value and right_rms_value.

[0120] (8) After removing the DC component, divide the total left_rms_value and right_rms_value by the number of frequency groups of the stft matrix to calculate the average rms values of the left and right microphone signals, and logarithmically convert them to decibels (dB).

[0121] (9) Calculate the dB difference db_diff between the RMS values of the left and right microphone signals.

[0122] (10) Calculate the linear ratio linearly using the dB difference, and then combine it with the pipe length to calculate the distance from the leak point to the left microphone, thus realizing positioning and ranging.

[0123] The overall logic is to analyze the audio signals collected by the left and right microphones, calculate their frequency domain characteristics (STFT), and perform anomaly processing and smoothing processing. Finally, convert the difference in the RMS values of the left and right microphone signals into an energy ratio and combine it with the pipe length to obtain the position of the leaking water point. And through pilot tests in various real scenarios and continuous optimization of the algorithm model, the accuracy of locating the leaking water point can currently reach 97.5%, and the error value is controlled within about 2 meters.

[0124] The implementation and deployment method adopts the NMA method to monitor water supply pipes, especially old pipes, in real time. Implement intelligent transformation on the water supply network, integrating IoT devices, machine learning, cloud platforms, big data, and mobile Internet technologies. By synchronously laying relevant sensors during the construction, transformation, and real-time monitoring of the water supply network, realize intelligent water supply management based on the Internet of Things, and can better solve the problem of water loss in the water supply industry. - One is to reduce losses, from "manual inspection" to "accurate positioning". Once a suspected leakage occurs in the pipe network, the intelligent noise leak detector can accurately determine the location of the leak point.

[0125] 4.1 Equipment Layout Requirements

[0126] Mainly lay out points for water supply pipes over 10 years old or pipes with frequent leakage warnings in the DMA. Take two unmanned inspection leak noise monitors as a group and install them in two pipe wells at intervals of 500 - 800 meters respectively.

[0127] 4.2 Preliminary Preparation

[0128] Survey pipes over 10 years old or areas with frequent leakage warnings in the DMA;

[0129] Study the survey results to determine the equipment layout method, layout scope, layout location, etc.;

[0130] According to the studied equipment layout plan, conduct on-site surveys to confirm the on-site point conditions, including the specific location of the pipe well, road conditions, etc.;

[0131] Check the hardware equipment. The inspection contents include whether the battery of the equipment is fully charged and whether the equipment is running normally.

[0132] 4.3 Leakage Identification and Repair Process

[0133] After installing the Internet of Things devices, the sound of the pipeline is collected, and then enters the leakage detection data model and the leakage location model for data conversion and comparison. The results are output to the maintenance work orders of the water utility company and the leakage points are displayed on the GIS map. The maintenance unit of the water utility company repairs according to the output results and returns the maintenance results to the intelligent water system.

[0134] The foregoing has shown and described the basic principles, main features and advantages of the present invention. For a person skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms. Therefore, in any regard, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, it is intended to embrace all changes falling within the meaning and scope of the equivalent elements of the claims in the present invention. Any reference signs in the claims should not be construed as limiting the claims concerned.

[0135] In addition, it should be understood that although this specification is described according to embodiments, not every embodiment only contains an independent technical solution. This narrative manner of the specification is only for clarity. A person skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by a person skilled in the art.

Claims

1. A method for artificial intelligence leakage identification and location of municipal water supply and drainage pipe networks, characterized in that, It includes the following steps: Step 1: Data collection: High-sensitivity sensors installed on pipes or valves are used to collect real-time audio data of the water flow in the pipes, and the data is transmitted to the data processing center through Internet of Things technology; Step 2: Leak detection data model analysis: Feature extraction and selection are performed on the collected audio data, and after on-site confirmation, it is marked to form a data set containing leakage data and non-leakage data; Step 3: Leak point location model establishment: Extract the features of the audio data, use a machine learning classifier for model training and cross-validation, select the best hyperparameters to establish a leak detection data model, design a custom function to process the audio data, through frequency domain feature analysis, anomaly processing and smoothing processing, combined with the pipe length to calculate the distance from the leak point to the sensor, and establish a leak point location model; Step 4: Alarm and work order generation: Once a water leakage signal is detected, the system automatically generates an alarm message and uploads it to the work order business system of the water utility company; Step 5: On-site confirmation and repair: The leak detection personnel and repair personnel of the water utility company arrive at the site according to the location information, reconfirm the water leakage situation, excavate near the location point, find and repair the actual water leakage point.

2. The method for artificial intelligence leakage discrimination and positioning of a municipal water supply and drainage pipe network according to claim 1, characterized in that: The sensitivity of the high-sensitivity sensor in Step 1 is not less than 1200 pC / g, and the sampling rate is 8192 Hz.

3. The method for artificial intelligence leakage identification and location of a municipal water supply and drainage pipe network according to claim 1, characterized in that: In Step 1, the data collection time is selected during the period with the least environmental interference.

4. The method for artificial intelligence leakage identification and positioning of a municipal water supply and drainage pipe network according to claim 1, characterized in that: The features extracted in Step 3 include linear time-domain features, frequency-domain features, and features describing the non-linear features of the leakage signal.

5. A method for artificial intelligence leakage identification and location of a municipal water supply and drainage pipe network according to claim 1, characterized in that: The machine learning classifier used in Step 3 is selected from at least one of support vector machine (SVM), decision tree (DT), K-nearest neighbor (KNN), extreme gradient boosting (XGBoost), and light gradient boosting (LightGBM).

6. A method for artificial intelligence leakage identification and location of a municipal water supply and drainage pipe network according to claim 1, characterized in that: The custom function in Step 3 includes short-time Fourier transform (STFT), smoothing processing, root mean square (RMS) calculation, and maximum correlation analysis (MRCA).

7. A method for artificial intelligence leakage identification and location of a municipal water supply and drainage pipe network according to claim 1, characterized in that: The alarm message in Step 4 contains information such as the detection device number, installation address, water leakage identification, and water leakage location.

8. A method for artificial intelligence leakage identification and location of a municipal water supply and drainage pipe network according to claim 1, characterized in that: After the repair in Step 5 is completed, the repair result is returned to the intelligent water service system to form a closed-loop management.

9. The method for artificial intelligence leakage identification and positioning of a municipal water supply and drainage pipe network according to claim 1, characterized in that: It also includes data analysis and optimization: Based on the data samples, deep learning methods are used to autonomously detect leakage signals to improve the accuracy of identification.

10. The method for artificial intelligence leakage identification and location of a municipal water supply and drainage pipe network according to claim 1, characterized in that: It also includes step system security and maintenance: Ensure the security of data transmission, storage and processing, and regularly maintain and upgrade the system to ensure the stable operation of the system and the accuracy of data.

Images