Digital twin system for natural gas leak detection and positioning in integrated pipeline corridor based on sound source positioning

Abstract

The present application relates to the field of digital twin technology, and provides a digital twin system for natural gas leak detection and positioning in an integrated pipe gallery based on sound source positioning. In the system, a methane sensor is arranged in a natural gas tank of an integrated pipe gallery, and when the methane concentration is detected to reach a preset threshold, an alarm signal is sent to a mobile unit; the mobile unit autonomously moves to the natural gas leakage area in the natural gas tank based on the alarm signal, collects the natural gas leakage acoustic wave signal, and locates the natural gas leakage source through a leakage acoustic wave difference algorithm, and feeds back the located natural gas leakage source position to the mobile unit; the mobile unit autonomously moves to the natural gas leakage source position, and sends the leakage source position to the digital twin system of the integrated pipe gallery, and the twin system updates the positioning result to the digital twin model in real time, and provides a virtual visualization display in the integrated pipe gallery, including the leakage source position and the mobile unit path, to achieve efficient and accurate leakage detection and response.

Description

Technical Field

[0001] The present application relates to the field of digital twin technology, and in particular to a digital twin system for natural gas leak detection and positioning in an integrated pipeline corridor based on sound source positioning. Background Art

[0002] As a form of intensive utilization of urban underground space, the integrated pipeline corridor concentrates municipal utility pipelines (such as electricity, communications, water supply and drainage, natural gas, etc.) in the same space to achieve resource sharing and maintain boundaries. However, once an accident occurs in the integrated pipeline corridor, especially a natural gas leak, the consequences are often extremely dangerous.

[0003] As a flammable and explosive gas, natural gas accumulates in limited underground space, which can easily cause explosions and fires, resulting in serious property losses and casualties. In addition, natural gas leakage may also cause environmental pollution and secondary disasters, further aggravating the scope and severity of the accident. Therefore, it is particularly important to effectively monitor the integrated pipeline corridor, especially to prevent and respond to natural gas leakage accidents.

[0004] In a comprehensive pipeline corridor, natural gas leakage may bring serious safety hazards. Most existing natural gas leakage detection methods rely on direct detection by gas sensors, but in a complex pipeline corridor environment, the detection range of a single sensor is limited and may be interfered by other factors, making it difficult to accurately locate the source of the leak.

[0005] Therefore, there is an urgent need to provide a technical solution to the above-mentioned deficiencies in the prior art. Summary of the invention

[0006] The purpose of this application is to provide a digital twin system for natural gas leak detection and positioning in an integrated pipeline corridor based on sound source positioning to solve or alleviate the problems existing in the above-mentioned prior art.

[0007] In order to achieve the above objectives, this application provides the following technical solutions:

[0008] The present application provides a digital twin system for natural gas leakage detection and positioning in an integrated pipe gallery based on sound source positioning, comprising: a methane sensor, a mobile unit and a signal acquisition and positioning unit; the methane sensor is arranged in a natural gas tank of the integrated pipe gallery, and is used to monitor the methane concentration in the natural gas tank in real time, and send an alarm signal when it is detected that the methane concentration in the natural gas tank reaches a preset threshold; the mobile unit receives the alarm signal sent by the methane sensor, and autonomously moves to the natural gas leakage area in the natural gas tank based on the alarm signal; the signal acquisition and positioning unit is equipped on the mobile unit, and is provided with a microphone array, which is used to collect natural gas leakage sound wave signals in the natural gas leakage area, and according to the natural gas leakage sound wave signals, based on a pre-constructed leakage source positioning model, the natural gas leakage source position in the natural gas leakage area is leak-located by a leakage sound wave difference algorithm, and the located natural gas leakage source position is fed back to the mobile unit, so that the mobile unit autonomously moves to the corresponding natural gas leakage source position, and sends the natural gas leakage source position to the data convergence unit of the integrated pipe gallery.

[0009] Preferably, the signal acquisition and positioning unit is further configured to process the collected natural gas leakage sound wave signal based on the wavelet transform Kalman filtering method to obtain frequency spectrum data of the natural gas leakage sound wave signal, and determine the location of the natural gas leakage source in the natural gas leakage area based on the leakage sound wave difference algorithm according to the obtained frequency spectrum data of the natural gas leakage sound wave signal.

[0010] Preferably, the signal acquisition and positioning unit is based on the formula:

[0011]

[0012] The natural gas leak acoustic signal Decompose into a time-frequency distribution of different frequencies that vary over time ; In the formula, For the natural gas leak acoustic signal The scaling factor when performing wavelet transform; For the natural gas leak acoustic signal The translation time when performing wavelet transform; is an integral variable, representing the acquisition time of the natural gas leakage acoustic wave signal;

[0013] For the natural gas leak acoustic signal According to the zoom scale and pan time The complex conjugate of the scaled and translated wavelet function.

[0014] Preferably, the signal positioning acquisition unit is configured according to the formula:

[0015]

[0016] Determine the microphone sensor The natural gas leakage acoustic wave signal collected at all times According to the translation time When performing wavelet transform at two different scales The leakage sound wave frequency component ;

[0017] In the formula, For the natural gas leak acoustic signal Scaling when performing wavelet transform The differential of The natural gas leak acoustic signal The time-frequency distribution of different frequencies that change over time obtained by decomposition;

[0018] According to the formula:

[0019]

[0020] Sure The leakage sound wave frequency component at time Scaling The state prediction value on ;

[0021] In the formula, for The leakage sound wave frequency component at time At zoom scale The state prediction value on ; for The frequency component of the leakage sound wave at time At zoom scale The estimated value of the state on ;

[0022] in, according to The frequency component of the leakage sound wave at time At zoom scale The state prediction value on The frequency component of the leakage sound wave at time Take the weighted average to get; For the microphone sensor The natural gas leakage acoustic wave signal collected at all times According to the translation time When performing wavelet transform at two different scales The leakage sound wave frequency component between.

[0023] Preferably, the signal positioning acquisition unit is further configured to: The state prediction value is as follows:

[0024]

[0025] The frequency component of the leakage sound wave Update; where, is the leakage sound wave frequency component Updated zoom scale The estimated value of the state on ; is the leakage sound wave frequency component At zoom scale Kalman gain on ; is the observation matrix; is the leakage sound wave frequency component At zoom scale The predicted value of the state.

[0026] Preferably, the signal positioning acquisition unit is further configured to determine any pair of microphone sensors in the microphone array based on the cross-correlation function. Frequency components of natural gas leakage sound waves The delay difference ;in, , are all positive integers, and according to the formula:

[0027]

[0028] Determine any pair of microphone sensors in the microphone array Frequency components of natural gas leakage sound waves The delay difference ; is the translation time when the wavelet transform is performed on the natural gas leakage acoustic wave signal; is the acquisition time of the natural gas leakage acoustic wave signal;

[0029] For the microphone sensor Frequency components of natural gas leakage sound waves The cross-correlation function.

[0030] Preferably, it also includes: constructing a physical simulation model of the integrated pipeline corridor according to the physical parameters of the integrated pipeline corridor; simulating the natural gas leakage sound wave signals at different natural gas leakage source locations in the integrated pipeline corridor based on the physical simulation model, and determining the time delay difference simulation data of the simulated signal reaching any pair of microphone simulation sensors in the physical simulation model, and constructing the training data set of the leakage source positioning model according to the simulated position of the natural gas leakage source in the physical simulation model and the corresponding time delay difference simulation data.

[0031] Preferably, the delay difference matrix generated by the delay difference simulation data As input features, the corresponding simulated location of the natural gas leakage source is used as an output label to train the leakage source positioning model based on the convolutional recurrent neural network architecture; wherein, is the number of time frames, are respectively the dimensions of the time delay difference matrix between a corresponding pair of microphone sensors.

[0032] Preferably, according to any pair of microphone sensors in the microphone array Frequency components of natural gas leakage sound waves The delay difference , determine the predicted location of the natural gas leakage source in the integrated pipeline corridor through the leakage source positioning model; and evaluate the leakage source positioning model according to the predicted location of the natural gas leakage source and the location of the natural gas leakage source determined by the leakage acoustic wave difference algorithm.

[0033] Preferably, the evaluation of the leakage source positioning model includes determining a detection evaluation index and a positioning evaluation index of the leakage source positioning model. ; Wherein, the detection evaluation index includes: the leakage source positioning model is predicted when and the error rate of the predicted location of the natural gas leakage source output by the leakage source positioning model ; According to the formula:

[0034]

[0035] Determining the error rate of the predicted location of the natural gas leak source output by the leak source location model ; In the formula, , is the number of simulations when simulating the natural gas leakage in the integrated pipe gallery by using the physical simulation model; For the The number of simulated locations of natural gas leakage sources successfully predicted by the leakage source location model in the natural gas leakage simulation; For the the number of simulated locations of natural gas leakage sources omitted by the leakage source location model in the natural gas leakage simulation; For the The number of simulated locations of natural gas leakage sources that are incorrectly predicted by the leakage source location model in the natural gas leakage simulation; For the The actual number of simulated locations of the natural gas leakage source in the natural gas leakage simulation;

[0036] The positioning evaluation index It includes: the predicted position of the natural gas leakage source output by the leakage source positioning model, the angle between the connecting line of the mobile unit and the natural gas leakage source position determined by the leakage acoustic wave difference algorithm, and the connecting line of the mobile unit. , and the accuracy of the match between the predicted location of the natural gas leakage source output by the leakage source positioning model per unit time and the location of the natural gas leakage source determined by the leakage acoustic wave difference algorithm ;

[0037] According to the formula:

[0038]

[0039] Determine the angle between the predicted position of the natural gas leakage source output by the leakage source positioning model and the connecting line of the mobile unit and the natural gas leakage source position determined by the leakage acoustic wave difference algorithm and the connecting line of the mobile unit ; and the accuracy of the natural gas leakage source prediction position output by the leakage source positioning model per unit time and the natural gas leakage source position determined by the leakage acoustic wave difference algorithm matching ;

[0040] In the formula, is the number of microphone sensors in the microphone array; For the During the first natural gas leakage simulation, the sound source arrival angle of the natural gas leakage sound wave simulation signal in the physical simulation model; For the During the first natural gas leakage simulation, the sound source arrival angle of the natural gas leakage sound wave simulation signal predicted by the leakage sound wave location model; For the During the first natural gas leakage simulation, the distance between the simulated position of the natural gas leakage source in the physical simulation model and the microphone simulation sensor corresponding to the microphone array; For the During the first natural gas leakage simulation, the distance between the predicted position of the natural gas leakage source predicted by the leakage sound source localization model and the microphone simulation sensor corresponding to the microphone array.

[0041] Beneficial effects:

[0042] In the digital twin system for natural gas leak detection and positioning in an integrated pipeline corridor based on sound source positioning provided in an embodiment of the present application, a methane sensor is arranged in a natural gas tank of the integrated pipeline corridor, and is used to monitor the methane concentration in the natural gas tank in real time, and when it is detected that the methane concentration in the natural gas tank reaches a preset threshold, an alarm signal is sent to a mobile unit; the mobile unit receives the alarm signal sent by the methane sensor, and autonomously moves to the natural gas leakage area in the natural gas tank based on the alarm signal; the mobile unit is equipped with a signal acquisition unit, and a microphone array is provided in the signal acquisition unit, which is used to collect natural gas leakage sound wave signals in the natural gas leakage area, and based on the natural gas leakage sound wave signals, the natural gas leakage source position in the natural gas leakage area is leak-located by a leakage sound wave difference algorithm, and the located natural gas leakage source position is fed back to the mobile unit; the mobile unit autonomously moves to the corresponding natural gas leakage source position, and sends the natural gas leakage source position to the data aggregation unit of the integrated pipeline corridor.

[0043] Therefore, by using the microphone array to monitor the acoustic wave signals of natural gas leakage, the source of natural gas leakage can be accurately located through the acoustic wave difference, overcoming the problems of limited detection range and susceptibility to interference of traditional gas sensors in complex environments, thereby achieving timely, rapid and accurate detection of natural gas leaks in the integrated pipeline corridor, providing support for the accurate positioning of the source of natural gas leakage, and effectively shortening the accident response time. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. Among them:

[0045] Figure 1 A schematic diagram of a scenario of a digital twin system for natural gas leak detection and positioning in an integrated pipeline corridor based on sound source positioning according to some embodiments of the present application;

[0046] Figure 2 A technical schematic diagram of a digital twin system for detecting and locating natural gas leaks in an integrated pipeline corridor based on sound source positioning according to some embodiments of the present application;

[0047] Figure 3 A schematic diagram of a process for natural gas leakage detection in an integrated pipeline gallery based on sound source positioning according to some embodiments of the present application;

[0048] Figure 4 A digital twin framework architecture design diagram provided according to some embodiments of the present application;

[0049] Figure 5 A two-dimensional spectrum diagram after wavelet transformation provided according to some embodiments of the present application.

[0050] Figure 6 A one-dimensional spectrum diagram after wavelet transformation according to some embodiments of the present application DETAILED DESCRIPTION

[0051] The present application will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments. Each example is provided by way of explanation of the present application and does not limit the present application. In fact, it will be clear to those skilled in the art that modifications and variations may be made in the present application without departing from the scope or spirit of the present application. For example, a feature shown or described as a part of an embodiment may be used in another embodiment to produce yet another embodiment. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the embodiments of the present invention should belong to the scope of protection of the embodiments of the present invention.

[0052] At present, traditional gas sensors are easily interfered with when monitoring natural gas leaks in complex pipelines, resulting in inaccurate positioning of the natural gas leakage source. In this application, a digital twin system for natural gas leak detection and positioning in an integrated pipeline corridor based on sound source positioning is proposed. With the help of a microphone array, the natural gas leak is accurately positioned through acoustic wave analysis technology, thereby achieving timely and accurate positioning of the natural gas leakage source in a complex pipeline corridor environment.

[0053] The digital twin system is designed in layers according to the physical layer, data layer, algorithm layer, and functional layer. The physical layer includes physical entities and logical rules. The physical entity is the actual pipeline corridor facilities and the natural gas transmission pipelines they contain, while the logical rules define the operation mode and behavior rules of these physical objects in the digital twin system. The logical layer ensures that the system can accurately map the physical state in reality, which serves as the basis of the digital twin system data.

[0054] The data layer is responsible for collecting and managing fixed data from physical space and real-time data collected by sensors, including environmental data in the pipeline corridor, methane sensor data, and leak detection data, to ensure that the algorithm layer can process and analyze based on accurate and timely data. In the algorithm layer, deep learning, signal processing, and noise reduction processing technologies are used to analyze and calculate the raw data provided by the data layer. Through the deep learning model, the characteristics of natural gas leaks are identified, and the accuracy of data analysis is improved by combining signal processing technology to achieve accurate positioning of the leak source.

[0055] The functional layer specifically includes leak alarm, visualization of pipeline corridor data, and location of leak sources. Through real-time monitoring and data analysis, the functional layer can visualize the status of natural gas in the pipeline corridor, issue an alarm when an abnormality is detected, and locate the leak for timely processing.

[0056] Specifically, the integrated pipe gallery server models the integrated pipe gallery experimental cabin and various types of sensors through blender to obtain virtual models of various objects on the experimental platform. Then, the virtual model is given material through the shader, and finally the texture map is obtained through baking, and the virtual model and the corresponding texture map are imported into Unity 3D. A digital twin system is established in Unity 3D, and the model position is set in Unity 3D based on the geometric relationship of various virtual models.

[0057] like Figures 1 to 6 As shown in the figure, the digital twin system for detecting and locating natural gas leaks in the integrated pipe gallery based on sound source positioning includes: a methane sensor, a mobile unit, and a signal acquisition and positioning unit. Among them, the methane sensor is arranged in the natural gas tank of the integrated pipe gallery, and according to the urban underground integrated pipe gallery pipeline project, a methane sensor is arranged every 15 meters to realize real-time monitoring of the methane concentration in the natural gas tank. When a natural gas leak occurs in the pipeline, the methane sensor detects that the methane concentration in the natural gas tank reaches the preset threshold, and sends a natural gas leak alarm signal.

[0058] The mobile unit receives the alarm signal sent by the methane sensor and moves autonomously to the natural gas leakage area in the natural gas tank based on the alarm signal. The mobile unit uses an autonomously navigated mobile robot that can move autonomously in the pipe gallery. Specifically, the mobile robot is equipped with a wireless communication module connected to the intranet of the pipe gallery. The alarm signal is quickly transmitted to the mobile unit through the intranet communication, and the alarm signal carries the number and coordinate information of the methane sensor that issued the alarm signal, indicating the area where the leakage may occur.

[0059] The mobile robot has a built-in laser radar for synchronous positioning, map construction and obstacle avoidance, and the corresponding driver package is installed and the laser radar node rplidar_ros is started. In the mobile robot, Raspberry Pi is used as the main control chip, communicating with ROS through the serial port, using rosserial or ros_control package to control the mobile robot chassis, and the linear speed and angular speed of the mobile robot are controlled by publishing cmd_vel in ROS, and the move_base method is used to provide the mobile robot with action-based path planning, and the path navigation of the mobile robot is realized through the global path planning and local real-time planning planner.

[0060] In addition, obstacles and path weights can be represented by importing two cost maps, gloable_costmap and local_planner, into the mobile robot, detecting the surrounding environment in real time through sensors, and dynamically adjusting the direction of travel to avoid obstacles. Set multiple cruise points, access these cruise points one by one through the ROS navigation stack, and write a ROS node that will send multiple target points to move_base in a loop. When the alarm signal is received, the mobile robot starts automatic navigation and quickly goes to the area where the leak may occur. It should be noted that when a natural gas leak occurs, multiple methane sensors may send alarm signals at the same time. According to the alarm signal, it can only be determined that a natural gas leak has occurred in the area where the methane sensor is located, but the exact location of the natural gas leak cannot be determined.

[0061] Based on this, a signal acquisition and positioning unit is equipped on the autonomous navigation mobile robot. When the mobile robot moves to the possible leakage area according to the alarm signal, it detects the sound waves generated by the natural gas leakage, collects the natural gas leakage sound wave signals, and locates the position of the natural gas leakage source through the sound wave difference. Specifically, a microphone array is set on the signal acquisition unit to collect the natural gas leakage sound wave signals in the natural gas leakage area. Here, the microphone array is arranged in a circular structure array to collect the natural gas leakage sound wave signals, so as to better improve the spatial resolution of the natural gas leakage sound wave signals, enhance the all-round perception ability of the natural gas leakage sound wave signals in the pipeline corridor, and enable it to effectively capture the sound wave signals transmitted in different directions in a complex environment.

[0062] In a specific example, the signal acquisition and positioning unit is configured to process the collected natural gas leakage sound wave signal based on the wavelet transform Kalman filter method to obtain the spectrum data of the natural gas leakage sound wave signal (i.e., the leakage sound wave frequency component). Specifically, according to the formula:

[0063]

[0064] Natural gas leak acoustic signal Decompose into a time-frequency distribution of different frequencies that vary over time In the formula, is an integral variable, representing the acquisition time of the natural gas leakage acoustic wave signal. To detect the acoustic signal of natural gas leak The scaling factor when performing wavelet transform, To detect the acoustic signal of natural gas leak The translation time when performing wavelet transform.

[0065] To detect the acoustic signal of natural gas leak According to the zoom scale and pan time The complex conjugate of the scaled and translated wavelet function.

[0066] Shift Time Represents the translation of the wavelet function on the time axis. By changing the translation time , the wavelet function can move left and right on the time axis, so as to analyze the natural gas leak acoustic signal at different time points. When increases, the wavelet function moves to the right, indicating the analysis of events that occurred later in the natural gas leak acoustic signal on the time axis; When it decreases, the wavelet function moves to the left, indicating on the time axis the analysis of events that occurred earlier in the natural gas leak acoustic signal.

[0067] Scaling Characterizes the width or degree of expansion of the wavelet function and determines the resolution of the wavelet function in the frequency domain. Corresponding to the expansion of the wavelet function, that is, the wavelet function becomes wider in the time domain, which can capture the low-frequency components of the natural gas leakage acoustic signal, that is, a larger scaling scale Corresponding to the low-frequency analysis of the acoustic signal of a natural gas leak. Smaller zoom scale Corresponding to the contraction of the wavelet function, that is, the wavelet function becomes narrower in the time domain, which can capture the high-frequency components of the natural gas leakage acoustic signal, that is, a larger scaling scale Corresponding to high-frequency analysis of natural gas leak acoustic signals.

[0068] By adjusting the scaling of the wavelet transform and pan time , decomposing the natural gas leakage acoustic wave signal into components of different frequencies and times, so that the leakage acoustic wave signal can be finely analyzed in time and frequency and useful signal components can be extracted from it.

[0069] After confirming the acoustic signal of natural gas leakage Time-frequency distribution of different frequencies changing over time Then, according to the formula:

[0070]

[0071] Identify the microphone sensor Natural gas leakage acoustic signals collected at all times According to usual time When performing wavelet transform at two different scales The leakage sound wave frequency component In the formula, To detect the acoustic signal of natural gas leak Scaling when performing wavelet transform The differential of .

[0072] Then, according to the formula:

[0073]

[0074] Sure The frequency component of the leakage sound wave at time Scaling The state prediction value on the leakage sound wave frequency component Make predictions. In the formula, is the leakage sound wave frequency component At zoom scale The state prediction value on ; is the leakage sound wave frequency component At zoom scale The estimated value of the state on is the state transfer matrix of the Kalman filter.

[0075] Here, the state prediction value is calculated based on the previous existing information, and the state estimation value is corrected based on the predicted value combined with the current observation data. The frequency component of the leakage sound wave at time At zoom scale The predicted value of the state and the measured value of the microphone sensor ( The frequency component of the leakage sound wave measured at any time ) and perform weighted averaging to obtain The frequency component of the leakage sound wave at time At zoom scale The estimated value of the state .

[0076] Among them, the weight of the weighted average is adjusted according to the Kalman gain; when the prediction error covariance matrix of the leakage sound wave frequency component Very small, indicating that the frequency component of the leakage sound wave is on the scale The state prediction value on is reliable, and the Kalman gain It tends to reduce the weight of the measured value of the microphone sensor and rely more on the frequency component of the leakage sound wave in the scaling scale. The state prediction value on ; when the covariance matrix of the observation noise When it is larger, it means that the measured value of the microphone sensor is more noisy (unreliable), and the Kalman gain This will also reduce the weight of the microphone sensor's measured value. On the contrary, if the measured value of the wind sensor is more reliable (i.e., the covariance matrix of the observed noise is smaller), Kalman gain The weight of the measured value of the wind sensor will be increased, so that the state estimation value is more dependent on the measured value of the current wind sensor.

[0077] Next, according to the leakage sound wave frequency component At zoom scale The state prediction value on , according to the formula:

[0078]

[0079] Frequency components of leakage sound waves To update. is the leakage sound wave frequency component Updated zoom scale The estimated value of the state on ; is the leakage sound wave frequency component At zoom scale Kalman gain on ; is the observation matrix.

[0080] After obtaining the spectrum data of the natural gas leakage sound wave signal, the location of the natural gas leakage source in the natural gas leakage area is determined based on the leakage sound wave difference algorithm. Specifically, any pair of microphone sensors in the microphone array is determined based on the cross-correlation function. Frequency components of natural gas leakage sound waves The delay difference Then, according to the microphone sensor Frequency components of natural gas leakage sound waves The delay difference , determine the location of the natural gas leak source within the natural gas leak area.

[0081] in, , are all positive integers, and according to the formula:

[0082]

[0083] Identify any pair of microphone sensors in the microphone array Frequency components of natural gas leakage sound waves The delay difference In the formula, is the translation time of the natural gas leakage acoustic signal during wavelet transformation, i.e., the microphone sensor Acoustic signal of natural gas leakage Relative to microphone sensor Collecting natural gas leak acoustic signals delay time. For microphone sensor Frequency components of natural gas leakage sound waves The cross-correlation function of The peak value is the microphone sensor Frequency components of natural gas leakage sound waves The delay difference .

[0084] In this application, the acoustic simulation tool Roomsimove can be used to set the propagation path of the natural gas leakage sound wave according to the physical parameters of the pipeline corridor (such as wall materials, dimensions, etc.), including multipath reflection and sound wave absorption characteristics; set natural gas leakage sound sources at different locations to generate acoustic signals emitted by different sound sources. The sound source signal includes pulse signals and noise, which are used to simulate the sound source in the actual environment. Specifically, according to the physical parameters of the integrated pipeline corridor, a physical simulation model of the integrated pipeline corridor is constructed to simulate the natural gas leakage sound wave signals at different natural gas leakage source locations in the integrated pipeline corridor.

[0085] Then, the delay difference simulation data of the simulated signal reaching any pair of microphone simulation sensors in the physical simulation model is determined, and the training data set of the leakage source positioning model is constructed according to the simulated position of the natural gas leakage source in the physical simulation model and the corresponding delay difference simulation data. Specifically, the arrival delay difference between the simulated signal reaching each pair of microphone simulation sensors is calculated, and the delay difference simulation data is recorded as the input feature of the leakage source positioning model training. The simulated position of the natural gas leakage source corresponding to each simulated signal is recorded as the output label of the leakage source positioning model training.

[0086] That is to say, the delay difference features and the corresponding sound source location labels are combined into a complete data set (i.e., training data set) for the training of the leakage source location model. Specifically, the delay difference matrix generated by the delay difference simulation data is As input features, the corresponding simulated location of the natural gas leak source is used as the output label to train the leak source location model based on the convolutional recurrent neural network architecture. is the number of time frames, are the dimensions of the delay difference matrix between the corresponding pair of microphone sensors. Characterizes the frequency of data collection per unit time; are the number of rows and columns of the delay difference matrix respectively.

[0087] In this application, data is processed based on a convolutional recurrent neural network architecture. Specifically, the delay difference simulation data is input into the input layer of the convolutional recurrent neural network architecture, and then the delay difference matrix is ​​extracted through multiple convolutional layers. Among them, each convolution layer uses several convolution kernels, and the size of the convolution kernel can be determined by tuning so as to extract the relevant features between different microphone simulation sensor pairs from the delay difference matrix.

[0088] The convolution layer is followed by a maximum pooling layer to reduce the data dimension and highlight the main features. Then, the feature sequence extracted by the convolution layer is passed to the recurrent neural network LSTM, and the temporal dependencies in the time series are captured through the recurrent layer. In this way, the LSTM unit can learn the association between multiple frames of data and further improve the positioning accuracy.

[0089] The output of the recursive layer is passed to the fully connected layer for further feature extraction and mapping, and the output of the convolution and recursive layers is converted into the spatial position of the sound source. The coordinate position of the sound source is finally output through the output layer. , which is the spatial position of the sound source.

[0090] In this application, the difference between the predicted location of the natural gas leakage source in the integrated pipeline corridor determined by the leakage source positioning model and the actual location (the location of the natural gas leakage source determined by the leakage acoustic wave difference algorithm) is calculated by the mean square error loss function. At the same time, the leakage source positioning model is evaluated through the determined detection evaluation indicators and positioning evaluation indicators.

[0091] Among them, the mean square error loss function is:

[0092]

[0093] In the formula, In order to conduct the first The coordinates of the simulated leakage location during the first natural gas leakage simulation; The leak source location model determines the The coordinates of the predicted location of the natural gas leakage source corresponding to the leakage simulation location during the natural gas leakage simulation; It is the number of simulations when simulating natural gas leakage in the integrated pipeline corridor through the physical simulation model.

[0094] According to the formula:

[0095]

[0096] Determine the detection and evaluation indicators of the leakage source location model.

[0097] In the formula, the leakage source location model is used for prediction It is used to measure the overall detection effect of the leakage source location model. It is the average value of precision and recall. The closer the value is to 1, the better the performance of the model is. It can accurately detect the leakage point (high precision) and find most leakages (high recall). The error rate of the predicted location of the natural gas leakage source output by the leakage source positioning model is the error rate of the predicted location of the natural gas leakage source in the integrated pipeline corridor through the physical simulation model. In the natural gas leakage simulation, the number of times the leakage source location model predicted errors accounted for The proportion of simulated natural gas leaks;

[0098] For the The number of simulated locations of natural gas leak sources successfully predicted by the leak source location model in each natural gas leak simulation. It should be noted that in each natural gas leak simulation, there may be multiple natural gas leak points at the same time. That is, the number of natural gas leakage points successfully predicted by the leakage source location model.

[0099] For the The number of simulated locations of natural gas leakage sources missed by the leakage source location model in the natural gas leakage simulation, that is, the number of natural gas leakage points not predicted by the leakage source location model. For the The number of simulated locations of natural gas leak sources that were incorrectly predicted by the leak source location model in the natural gas leak simulation; that is, the number of incorrectly predicted natural gas leak points. For the The number of actual leakage simulation locations in the natural gas leakage simulation, that is, the total number of leakage simulation points in the physical simulation model. , It is the number of simulations when simulating natural gas leakage in the integrated pipeline corridor through the physical simulation model.

[0100] is a true positive sample, that is In the first natural gas leak simulation, the leakage source location model predicts the number of correct samples (the number of predicted locations of the correct natural gas leak sources); that is, in the In the natural gas leakage simulation, the leakage source location model predicted a positive value and the actual value was positive, that is, the predicted location of the natural gas leakage source obtained by the leakage source location model was the same as the location of the natural gas leakage source determined by the leakage acoustic wave difference algorithm. For a false positive sample, the leak source location model predicts it to be positive but it is actually negative, that is, the predicted location of the natural gas leak source obtained by the leak source location model is different from the location of the natural gas leak source determined by the leak sound wave difference algorithm; It represents a false negative sample, which is predicted by the leakage source location model as negative but is actually positive. That is, the leakage source location model does not obtain the predicted location of the natural gas leakage source, but the leakage sound wave difference algorithm determines that there is a natural gas leakage source location.

[0101] According to the formula:

[0102]

[0103] Determine the location evaluation index of the leakage source location model In the formula, The angle between the predicted position of the natural gas leakage source and the connecting line of the mobile unit output by the leakage source positioning model and the natural gas leakage source position and the connecting line of the mobile unit determined by the leakage acoustic wave difference algorithm; It is the accuracy rate of matching the predicted position of the natural gas leakage source output by the leakage source positioning model per unit time with the natural gas leakage source position determined by the leakage acoustic wave difference algorithm.

[0104] is the number of microphone sensors in the microphone array; For the The sound source arrival angle of the natural gas leakage sound wave simulation signal in the physical simulation model during the first natural gas leakage simulation; For the During the first natural gas leak simulation, the sound source arrival angle of the natural gas leak sound wave simulation signal predicted by the leak source location model; For the During the first natural gas leak simulation, the distance between the simulated position of the natural gas leak source in the physical simulation model and the microphone simulation sensor corresponding to the microphone array; For the During the natural gas leakage simulation, the distance between the predicted position of the natural gas leakage source predicted by the leakage source positioning model and the microphone simulation sensor corresponding to the microphone array; It is the Hungarian algorithm, which is used to measure the difference between the sound source arrival angle of the natural gas leak sound wave simulation signal in the physical simulation model and the sound source arrival angle of the natural gas leak sound wave simulation signal predicted by the leakage source location model.

[0105] Generally, the closer ER is to 0, FR is to 1, F1-Score is to 1, and DE is to 0, the better the performance of the leakage source location model. The leakage source location model is continuously updated through iterations, and the prediction accuracy is continuously improved. In the integrated pipe gallery, after the microphone sensor is determined, Natural gas leak acoustic signal The delay difference After that, the leakage source location model can be used to accurately locate the natural gas leakage source in the natural gas leakage area.

[0106] In each integrated pipeline corridor, there is a corresponding data aggregation unit and a fire alarm unit. When the signal acquisition and positioning unit locates the natural gas leakage source in the pipeline corridor, the located natural gas leakage source position is fed back to the mobile unit. The mobile unit moves autonomously to the corresponding natural gas leakage source position and sends the natural gas leakage source position to the data aggregation unit and fire alarm unit of the integrated pipeline corridor so that timely response measures can be taken.

[0107] In this application, data transmission is performed through a Bluetooth module to ensure the stability, real-time performance and strong anti-interference performance of data transmission in a complex pipe gallery environment. Once the location of the natural gas leak source is determined, the location information and natural gas sensor data are quickly transmitted to the data aggregation unit of the integrated pipe gallery, that is, the control center server, through the Bluetooth module and recorded in the MySQL database, which not only ensures the integrity and accuracy of the data, but also greatly shortens the response time of accident handling and reduces the possibility of disasters.

[0108] All sensor data collected in the tunnel will be transmitted to the data aggregation unit, stored in the MySQL database, and then transmitted to the digital twin system. The digital twin system dynamically constructs the location of the leakage source by acquiring and processing this data in real time. In this process, the virtual model of the digital twin system is highly synchronized with the physical equipment. When the location of the leakage source in the physical system changes, the virtual model will be updated in real time to ensure consistency in the prediction and monitoring of the leakage source.

[0109] The predicted leak source location is visualized through the Unity 3D platform. The data transmitted via Bluetooth is connected to Unity 3D. The real-time Bluetooth data access is achieved by installing InTheHand.Bluetooth through the NuGet package manager, and then the SerialPort class is used to communicate with the Bluetooth device. The data (the location of the natural gas leak source) transmitted back by Bluetooth is parsed and the corresponding mobile robot position in Unity is updated.

[0110] In the Canvas tool, drag and drop the UI-Text object into the positionText field to display the position data of the mobile robot in real time. That is, use the UI-Text component in the Canvas tool to update the coordinate information of the leak source in real time to ensure that the system operator can clearly see the location and movement trajectory of the leak source in three-dimensional space. By combining the Xcharts graphic component with the MySQL database, the digital twin system can display the data in the corridor in the form of a chart. By calling Mysql.Connector and Xcharts API, the leakage data in the database can be visualized.

[0111] Build various digital charts in Unity 3D through the Xcharts plug-in, access database data through dynamic links, import Mysql.Data.dll into the Unity Assets folder, and then create a C# script. Connect the chart data source to MySQL real-time data through the API provided by MySQL Connector and Xcharts, query SQL through command.CommandText, obtain library information through the render.Read() method, and finally change the chart data source to real-time data in the MySQL library through data.Add (serieData) to form a natural gas detection digital twin system.

[0112] In the description of the present invention, the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0113] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A digital twin system for natural gas leak detection and positioning in an integrated pipeline corridor based on sound source positioning, characterized in that: include: Methane sensor, mobile unit and signal acquisition and positioning unit; The methane sensor is arranged in the natural gas tank of the integrated pipe gallery, and is used to monitor the methane concentration in the natural gas tank in real time, and send an alarm signal when the methane concentration in the natural gas tank reaches a preset threshold; The mobile unit receives the alarm signal sent by the methane sensor and autonomously moves to the natural gas leakage area in the natural gas tank based on the alarm signal; The signal collection and positioning unit is equipped on the mobile unit and is provided with a microphone array. The microphone array is used to collect natural gas leakage sound wave signals in the natural gas leakage area, and based on the natural gas leakage sound wave signals and a pre-constructed leakage source positioning model, the leakage source position of the natural gas leakage in the natural gas leakage area is located by a leakage sound wave difference algorithm, and the located natural gas leakage source position is fed back to the mobile unit, so that the mobile unit moves autonomously to the corresponding natural gas leakage source position, and the natural gas leakage source position is sent to the data convergence unit of the integrated pipe gallery; The signal acquisition and positioning unit is further configured to process the collected natural gas leakage sound wave signal based on the wavelet transform Kalman filter method to obtain the spectrum data of the natural gas leakage sound wave signal, and determine the location of the natural gas leakage source in the natural gas leakage area based on the leakage sound wave difference algorithm according to the obtained spectrum data of the natural gas leakage sound wave signal; And the signal acquisition and positioning unit is according to the formula: ； The natural gas leak acoustic signal Decompose into a time-frequency distribution of different frequencies that vary over time ; In the formula, For the natural gas leak acoustic signal The scaling factor when performing wavelet transform; For the natural gas leak acoustic signal The translation time when performing wavelet transform; is an integral variable, representing the acquisition time of the natural gas leakage acoustic wave signal; For the natural gas leak acoustic signal According to the zoom scale and pan time The complex conjugate of the scaled and translated wavelet function; The signal positioning acquisition unit is according to the formula: ； Identify microphone sensors in the microphone array The natural gas leakage acoustic wave signal collected at all times According to the translation time When performing wavelet transform at two different scales The leakage sound wave frequency component ; In the formula, For the natural gas leak acoustic signal Scaling when performing wavelet transform The differential of The natural gas leak acoustic signal The time-frequency distribution of different frequencies that change over time obtained by decomposition; According to the formula: ； Sure The leakage sound wave frequency component at time Scaling The state prediction value on ; In the formula, for The leakage sound wave frequency component at time At zoom scale The state prediction value on ; for The frequency component of the leakage sound wave at time At zoom scale The estimated value of the state on ; in, according to The frequency component of the leakage sound wave at time At zoom scale The state prediction value on The frequency component of the leakage sound wave at time Take the weighted average to get; A microphone sensor in the microphone array The natural gas leakage acoustic wave signal collected at all times According to the translation time When performing wavelet transform at two different scales The leakage sound wave frequency component between The signal positioning acquisition unit locates the leakage sound wave frequency component according to the leakage sound wave frequency component. The state prediction value is as follows: ； The frequency component of the leakage sound wave Make updates; In the formula, is the leakage sound wave frequency component Updated zoom scale The estimated value of the state on ; is the leakage sound wave frequency component At zoom scale Kalman gain on ; is the observation matrix; is the leakage sound wave frequency component At zoom scale The state prediction value on ; The signal positioning acquisition unit is further configured to determine any pair of microphone sensors in the microphone array based on the cross-correlation function. Frequency components of natural gas leakage sound waves The delay difference ;in, , are all positive integers, and according to the formula: ； Determine any pair of microphone sensors in the microphone array Frequency components of natural gas leakage sound waves The delay difference ; is the translation time when the wavelet transform is performed on the natural gas leakage acoustic wave signal; is the acquisition time of the natural gas leakage acoustic wave signal; For the microphone sensor Frequency components of natural gas leakage sound waves The cross-correlation function.

2. According to the digital twin system for natural gas leak detection and positioning in a comprehensive pipeline corridor based on sound source positioning according to claim 1, it is characterized in that: Also includes: A physical simulation model of the integrated pipe gallery is constructed according to the physical parameters of the integrated pipe gallery; Based on the physical simulation model, the natural gas leakage sound wave signals at different natural gas leakage source locations in the integrated pipeline corridor are simulated, and the time delay difference simulation data of the simulated signal reaching any pair of microphone simulation sensors in the physical simulation model is determined. According to the simulated position of the natural gas leakage source in the physical simulation model and the corresponding time delay difference simulation data, a training data set of the leakage source positioning model is constructed.

3. The digital twin system for natural gas leak detection and positioning in a comprehensive pipeline corridor based on sound source positioning according to claim 2 is characterized in that: The delay difference matrix generated by the delay difference simulation data As input features, the corresponding simulated location of the natural gas leakage source is used as an output label to train the leakage source positioning model based on the convolutional recurrent neural network architecture; in, is the number of time frames, are respectively the dimensions of the time delay difference matrix between a corresponding pair of microphone sensors in the microphone array.

4. The digital twin system for natural gas leak detection and positioning in a comprehensive pipeline corridor based on sound source positioning according to claim 3 is characterized in that: According to any pair of microphone sensors in the microphone array Frequency components of natural gas leakage sound waves The delay difference , determining the predicted location of the natural gas leakage source in the integrated pipeline corridor through the leakage source positioning model; The leakage source positioning model is evaluated based on the predicted position of the natural gas leakage source and the position of the natural gas leakage source determined by the leakage sound wave difference algorithm.

5. The digital twin system for detecting and locating natural gas leaks in a comprehensive pipeline corridor based on sound source positioning according to claim 4 is characterized in that: The evaluation of the leakage source location model comprises: Determine the detection evaluation index and positioning evaluation index of the leakage source positioning model ; Wherein, the detection evaluation index includes: the leakage source positioning model is predicted when and the error rate of the predicted location of the natural gas leakage source output by the leakage source positioning model ; According to the formula: ； Determining the error rate of the predicted location of the natural gas leak source output by the leak source location model ; In the formula, , is the number of simulations when simulating the natural gas leakage in the integrated pipe gallery by using the physical simulation model; For the The number of simulated locations of natural gas leakage sources successfully predicted by the leakage source location model in the natural gas leakage simulation; For the the number of simulated locations of natural gas leakage sources omitted by the leakage source location model in the natural gas leakage simulation; For the The number of simulated locations of natural gas leakage sources that are incorrectly predicted by the leakage source location model in the natural gas leakage simulation; For the The actual number of simulated locations of the natural gas leakage source in the natural gas leakage simulation; The positioning evaluation index It includes: the predicted position of the natural gas leakage source output by the leakage source positioning model, the angle between the connecting line of the mobile unit and the natural gas leakage source position determined by the leakage acoustic wave difference algorithm, and the connecting line of the mobile unit. , and the accuracy of the match between the predicted location of the natural gas leakage source output by the leakage source positioning model per unit time and the location of the natural gas leakage source determined by the leakage acoustic wave difference algorithm ; According to the formula: ； Determine the angle between the predicted position of the natural gas leakage source output by the leakage source positioning model and the connecting line of the mobile unit and the natural gas leakage source position determined by the leakage acoustic wave difference algorithm and the connecting line of the mobile unit ; and the accuracy of the natural gas leakage source prediction position output by the leakage source positioning model per unit time and the natural gas leakage source position determined by the leakage acoustic wave difference algorithm matching ; In the formula, is the number of microphone sensors in the microphone array; For the During the first natural gas leakage simulation, the sound source arrival angle of the natural gas leakage sound wave simulation signal in the physical simulation model; For the During the first natural gas leakage simulation, the sound source arrival angle of the natural gas leakage sound wave simulation signal predicted by the leakage sound wave location model; For the During the first natural gas leakage simulation, the distance between the simulated position of the natural gas leakage source in the physical simulation model and the microphone simulation sensor corresponding to the microphone array; For the During the natural gas leakage simulation, the distance between the predicted position of the natural gas leakage source predicted by the leakage source positioning model and the microphone simulation sensor corresponding to the microphone array.

Images