Urban pipeline leakage parameter-free positioning method and device

Abstract

The invention discloses a non-parameter positioning method and device for urban pipeline leakage. The method comprises the steps that first, second and third leakage signals corresponding to first, second and third sensors deployed on the two sides of a pipeline leakage point are acquired at the same time; obtaining a noise-reduced leakage signal through a pre-trained urban environment noise reduction model based on a complex convolutional network; performing cross-correlation calculation on the first and second leakage signals and the first and third leakage signals after noise reduction processing to obtain a time delay value; performing communication time delay estimation on the first sensor, the second sensor, the first sensor and the third sensor to obtain a time delay value after the communication time delay is compensated; obtaining the distance between the first sensor and the second sensor and the distance between the second sensor and the third sensor; calculating an accurate pipeline leakage position according to the obtained distance and the time delay value; according to the method, no parameter needs to be input, and high-precision pipeline leakage point positioning can be carried out in a noisy urban environment.

Description

Technical Field

[0001] The present invention relates to the technical field of urban pipeline leakage monitoring, and particularly to a method and device for parameterless positioning of urban pipeline leakage. Background Art

[0002] The rupture and damage of pipelines will lead to leakage, causing economic losses and huge waste of resources. Therefore, after a leakage occurs, it is very important to detect and locate the leakage point in time to repair the damaged pipeline for the stable operation of the urban pipeline system. Currently, the conventional leakage positioning method is the two-point time-delay correlation method. The sensors required by this method often need to be limitedly connected to avoid the influence of wireless communication delay, which limits the measurement range and is very troublesome to install at the same time. This method also requires manual input of the distance information between two points and the signal propagation speed of the pipeline. The distance information over hundreds of meters is often measured manually, with complex operations and low accuracy; the accuracy of the pipeline signal propagation speed is particularly crucial, but the measurement is more difficult. Usually, approximate values are selected by looking up tables based on known quantities such as pipe materials and pipe diameters. However, in practice, due to inevitable differences in pipelines produced by different countries and manufacturers, different landfill media and landfill depths of pipelines will also affect the measured speed value, resulting in relatively large measurement errors. At the same time, this method is very sensitive to noise interference. Especially in a noisy urban environment, it will cause certain time-delay measurement errors. If the signal-to-noise ratio of the measured signal is too low, it will even cause this method to completely fail. Existing noise reduction methods often require manual setting of thresholds and are difficult to handle the situation where the leakage signal and the noise signal have the same frequency, resulting in poor noise reduction effects in complex urban environmental noise. Summary of the Invention

[0003] Aiming at the above problems, the present invention provides a method and device for parameterless positioning of urban pipeline leakage, which can improve the convenience and accuracy of urban pipeline leakage positioning without manual input of parameters.

[0004] To achieve the above object, in a first aspect, the present invention provides a method for parameterless positioning of urban pipeline leakage, including the following steps:

[0005] Try to ensure that the first leakage signal, the second leakage signal, and the third leakage signal corresponding to the first sensor, the second sensor, and the third sensor deployed on both sides of the pipeline leakage point are obtained simultaneously. Among them, the first sensor, the second sensor, and the third sensor are distributed on the pipeline to be measured, the pipeline leakage point should be located between the first sensor and the second sensor, the third sensor should be close to the second sensor and far from the first sensor;

[0006] Through the pre-trained noise reduction model based on complex convolutional network for urban environment, the noise-reduced leakage signal No. 1, leakage signal No. 2, and leakage signal No. 3 are obtained;

[0007] performing cross-correlation calculations on the first leakage signal, the second leakage signal, the first leakage signal, and the third leakage signal after the noise reduction processing to obtain a time delay value;

[0008] Estimating communication delays of the first sensor, the second sensor, the first sensor, and the third sensor to obtain a time delay value after compensating for the communication delays;

[0009] Obtaining the distances between the first sensor, the second sensor, the second sensor, and the third sensor;

[0010] The precise location of the pipeline leakage point is calculated based on the obtained distance and the time delay value after compensating for the communication delay.

[0011] Accordingly, the steps of ensuring that the No. 1 leakage signal, No. 2 leakage signal, and No. 3 leakage signal corresponding to the No. 1 sensor, No. 2 sensor, and No. 3 sensor deployed on both sides of the pipeline leakage point are obtained simultaneously as much as possible include:

[0012] The first sensor is a master sensor, and the second and third sensors are slave sensors. Before the first master sensor starts to acquire the pipeline leakage point signal, it notifies the second and third slave sensors to start acquiring the pipeline leakage signal through wireless communication.

[0013] The host sensor No. 1 starts to obtain the pipeline leakage point signal;

[0014] After receiving the start acquisition signal sent by the first sensor, the second and third slave sensors immediately start acquiring the pipeline leakage point signal;

[0015] The obtained leakage signal No. 1, leakage signal No. 2 and leakage signal No. 3 are respectively clipped to ensure that the signal length is uniformly 4 seconds, the part exceeding 4 seconds is discarded, and the part less than 4 seconds is padded with zeros.

[0016] Optionally, the step of obtaining the noise-reduced first leakage signal, second leakage signal, and third leakage signal by using a pre-trained noise reduction model for urban environments based on a complex convolutional network includes:

[0017] Performing short-time Fourier transform on the first leakage signal, the second leakage signal, and the third leakage signal respectively to obtain time-frequency characteristic spectrograms of the leakage signals to be denoised;

[0018] Normalize the time-frequency feature spectrogram of the leakage signal to be noise-reduced;

[0019] Perform operations on the normalized time-frequency feature spectrogram using 7 encoder modules to obtain a compressed complex multi-dimensional feature vector;

[0020] Perform operations on the compressed complex multi-dimensional feature vector using 7 decoder modules, plus the output of the corresponding encoder module with skip connection, to obtain a noise-reduced complex ratio mask;

[0021] Perform a dot product operation on the noise-reduced complex ratio mask and the normalized time-frequency feature spectrogram to obtain a noise-reduced two-dimensional time-frequency feature spectrogram.

[0022] Perform an inverse short-time Fourier transform on the noise-reduced two-dimensional time-frequency feature spectrogram to obtain the noise-reduced acoustic signal;

[0023] Optionally, the encoder module, decoder module, and skip connection include:

[0024] The structures of the encoder module and decoder module are such that the input layer is a complex number composed of the real and imaginary parts of the feature spectrogram or a complex multi-dimensional vector feature output by the previous layer, the first layer is a complex two-dimensional convolution module, the second layer is a complex batch normalization module, the third layer is a complex LeakyRelu module, and the output layer is a complex multi-dimensional vector feature.

[0025] The skip connection is an adder structure connecting the encoder and the decoder, enabling the real and imaginary parts of the complex multi-dimensional feature vector to perform addition operations respectively.

[0026] Optionally, before obtaining the noise-reduced first leakage signal, second leakage signal, and third leakage signal through a pre-trained noise reduction model based on a complex convolution network for urban environments, it further includes training the complex two-dimensional convolution module network within the encoder and decoder modules, including:

[0027] Obtain training samples, where the training samples are several leakage acoustic signals and urban noise signals;

[0028] The input of the training set of the model is a noisy leakage acoustic signal formed by superimposing the leakage acoustic signal and the urban noise signal in the time domain with a random signal-to-noise ratio of (-5, 15) dB in the time domain;

[0029] The output of the training set of the model is the leakage acoustic signal corresponding to the input of the model training;

[0030] Automatically generate the corresponding initial weight matrix according to the initialization of the encoder and decoder modules.<�

[0031] Calculate and generate an independent audio source separation signal-to-noise ratio loss function corresponding to the encoder and decoder modules according to the training set of the model;

[0032] Backpropagate in the convolutional network according to the loss function, and update the weight matrix corresponding to the encoder-decoder module in combination with the gradient descent method until the loss function meets the preset conditions, and then generate the noise reduction network model.

[0033] Correspondingly, the steps for estimating the communication delay between the first sensor, the second sensor, the first sensor, and the third sensor include:

[0034] The first host sensor records its local time as t1 and immediately sends a first handshake signal to the slave sensor;

[0035] After receiving the first handshake signal, the slave sensor records its local time as t2;

[0036] The slave sensor records its local time as t3 and immediately sends a second handshake signal to the first host sensor;

[0037] After receiving the second handshake signal, the host sensor records its local time as t4;

[0038] Calculate and solve the communication delay τ according to the following formula:

[0039] τ = ((t4 - t3) + (t2 - t1)) / 2

[0040] Correspondingly, the steps for obtaining the distances between the first sensor, the second sensor, and the second sensor, the third sensor include:

[0041] Obtain the longitude and latitude information of the first sensor, the second sensor, and the third sensor;

[0042] Convert the longitude and latitude information into radian representation;

[0043] Apply the Mercator projection method to calculate the corresponding plane coordinates of each sensor;

[0044] Calculate the Euclidean distances between the plane coordinates of the first sensor, the second sensor, and the second sensor, the third sensor respectively.

[0045] Correspondingly, the steps for calculating the exact location of the pipeline leakage point according to the obtained distance and the time delay value include:

[0046] Define the time delay between the first leakage signal and the second leakage signal as Δt1, and the time delay between the first leakage signal and the third leakage signal as Δt2;

[0047] Define the communication time delay between the first sensor and the second sensor as τ1, and the time delay between the first sensor and the third sensor as τ2;

[0048] Define the distance between the first sensor and the second sensor as L1, and the distance between the second sensor and the third sensor as L2;

[0049] The distance x from the first main sensor to the pipeline leakage point, that is, the exact position of the pipeline leakage point, can be calculated by the following formula:

[0050]

[0051] In a second aspect, the present invention provides an urban pipeline leakage parameterless positioning device, including:

[0052] A control and calculation center device, equipped with a touchable display and a wireless communication interface, running a visual user interface program, which can be controlled by the user for the progress of the leakage positioning method, communicate with the signal acquisition device, store the data sent by the leakage signal acquisition device, and has high-performance computing capabilities, and can quickly implement algorithms such as noise reduction model calculation, correlation time delay method calculation, and leakage point calculation.

[0053] A signal acquisition device, to which the first sensor, the second sensor, and the third sensor all belong, is equipped with a signal acquisition device and a storage device to collect and store leakage signals, and perform clipping and zero-padding calculations. It is equipped with a wireless communication interface and a real-time operating system for accurate communication time delay estimation, and is simultaneously controlled by the control and calculation center device through wireless communication and can perform data transmission with it.

[0054] Compared with the existing technologies, the beneficial effects of the present invention are as follows:

[0055] 1) Adopting a wireless communication method, it is more convenient to use than traditional wired communication, can measure a farther range, and introduces a communication time delay estimation algorithm to solve the time delay problem of wireless communication.

[0056] 2) Using a dedicated noise reduction network for the urban environment for noise reduction preprocessing, compared with traditional noise reduction methods, it has better noise reduction performance and does not require setting manual parameters, effectively improving the signal-to-noise ratio of the leakage signal and the positioning accuracy of the subsequent positioning algorithm.

[0057] 3) The present invention does not require any manual parameter setting. It automatically obtains distance information using a position sensor and eliminates the information on the signal propagation speed of the pipeline using the three-point positioning method. While effectively improving the convenience of use, it eliminates the influence caused by parameter errors and effectively improves the positioning accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 FIG. is a schematic structural diagram of a signal acquisition device of an urban pipeline leakage parameter-free positioning device for the hardware operating environment related to the embodiment solution of the present invention;

[0059] Figure 2 FIG. is a schematic structural diagram of a control and calculation center device of an urban pipeline leakage parameter-free positioning device for the hardware operating environment related to the embodiment solution of the present invention;

[0060] Figure 3 FIG. is a schematic flow chart of the first embodiment of a method for parameter-free positioning of urban pipeline leakage according to the present invention;

[0061] Figure 4 FIG. is a specific structure of a pre-trained urban environment noise reduction model based on a complex convolutional network provided by the embodiment solution of the present invention;

[0062] Figure 5 FIG. is a specific structure of a complex encoder and decoder provided by the embodiment solution of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0063] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings of the present invention. It should be understood that the specific embodiments described herein are only for explaining the present invention and not for limiting the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0064] Refer to Figure 1 , Figure 1 FIG. is a schematic structural diagram of a signal acquisition device of an urban pipeline leakage parameter-free positioning device for the hardware operating environment related to the embodiment solution of the present invention.

[0065] As Figure 1 shown, the signal acquisition device may include a processor 21, such as a central processing unit (CPU), a wireless communication module 22, such as a communication module using radio frequency (RF) technology, a signal acquisition module 23, such as an audio signal acquisition module, a position acquisition module 24, such as a module using the global positioning system (GPS), a user interface module 25, a bus 26, and a memory 27.

[0066] Among them, the bus 26 is used to implement connections and communications between components. The memory 27 includes various media that can store program codes, such as read-only memory (ROM), random access memory (RAM), erasable programmable read-only memory (EPROM), magnetic disks, or optical discs, and is responsible for storing a real-time operating system (RTOS), leakage signals, and program codes of each module. This system requires the support of a real-time operating system to ensure the timeliness of the execution of the communication delay estimation program and reduce the error of communication delay estimation. The position acquisition module 24 should be able to acquire longitude and latitude information.

[0067] Refer to Figure 2 , Figure 2 which is a schematic structural diagram of the control and calculation center device of the urban pipeline leakage parameterless positioning device for the hardware operating environment involved in the solution of the embodiment of the present invention.

[0068] As Figure 2 shown, the control and calculation center device may include a high-performance processor 31, a wireless communication module 32, a touchable display screen 33, a bus 34, and a memory 35.

[0069] Among them, the high-performance processor 31 includes a central processing unit and one or more graphics processing units (GPUs). The central processing unit should include multiple cores. Optionally, a processor group may be composed of multiple central processing units and graphics processing units to meet the requirements of high-performance numerical calculations and neural network inferences. Optionally, the processor may also be other types of processors, etc., which are not limited in the embodiments of the present invention. The wireless communication module 32 should be of the same model as the wireless communication module 22 in the signal acquisition device. The bus 34 is used for connections and communications between components and is optionally used for mutual coupling of multiple processors between processor groups.

[0070] The memory 35 should have the same hardware structure as the memory 27 in the signal acquisition device, but requires a larger storage space to store more data. The memory 35 is responsible for storing an operating system, a visual user interface, and noise reduction model algorithms, related time delay algorithms, and leakage point calculation algorithm programs. Among them, the operating system does not have real-time requirements. The visual user interface and the touchable display screen 33 together implement user interaction functions, including process control, signal acquisition device management, data and result display, etc.

[0071] The embodiment of the present invention provides a method for parameterless positioning of urban pipeline leakage. Refer toFigure 3 , Figure 3 is a schematic flowchart of the first embodiment of a method for non-parameterized positioning of urban pipeline leaks according to the present invention.

[0072] In this embodiment, the method for non-parameterized positioning of urban pipeline leaks includes the following steps:

[0073] Step S10: Simultaneously obtain the first, second, and third leakage signals corresponding to the first, second, and third sensors deployed on both sides of the pipeline leakage point, respectively.

[0074] It should be noted that the first sensor, the second sensor, and the third sensor are distributed on the pipeline to be measured. The suspected leakage point of the pipeline should be located between the first sensor and the second sensor. The third sensor should be close to the second sensor and far from the first sensor. The signals collected by the first sensor, the second sensor, and the third sensor correspond to the first leakage signal, denoted as y1(t), the second leakage signal, denoted as y2(t), and the third leakage signal, denoted as y3(t).

[0075] It is easy to understand that the step of simultaneously obtaining the first leakage signal, the second leakage signal, and the third leakage signal corresponding to the first sensor, the second sensor, and the third sensor deployed on both sides of the pipeline leakage point as much as possible includes: before the first host sensor starts to obtain the pipeline leakage point signal, notify the second and third slave sensors to start obtaining the pipeline leakage signal through wireless communication; the first host sensor starts to obtain the pipeline leakage point signal; after receiving the start signal sent by the first sensor, the second and third slave sensors immediately start to obtain the pipeline leakage point signal; respectively perform clipping processing on the obtained first leakage signal, second leakage signal, and third leakage signal to ensure that the signal length is uniformly 4 seconds, and the part exceeding 4 seconds is discarded, and the part less than 4 seconds is filled with zeros. Optionally, the main bodies of the first sensor, the second sensor, and the third sensor can all adopt the aforementioned signal acquisition device, and other devices that can achieve the same or similar functions can also be used. This embodiment does not limit this.

[0076] Step S20: Obtain the denoised leakage signal through a pre-trained urban environment noise reduction model based on a complex convolutional network.

[0077] It should be noted that assume that the nature of the noise is additive noise, that is, the interference of the noise on the signal is manifested as the addition of the two in the time domain. Assume that the required clean signals are s1(t), s2(t), s3(t) respectively, and the mixed noise signals are n1(t), n2(t), n3(t) respectively, including various sources of noise and instrument noise in the city. The measured noisy signals are:

[0078] After denoising the signal \(y(t)=s(t)+n(t)\), an estimated signal of the clean signal is obtained. It is expected that the error between it and the clean signal \(s(t)\) is minimized. Considering the situation where the output of the noise reduction model is used for manual diagnosis by staff, the Scale Invariant Source to Noise Ratio (SISNR) is adopted as the optimization objective of the model loss function. The definition of SISNR is given as follows:

[0079]

[0080] The noise reduction problem is transformed into a supervised deep learning problem. The noisy signal as the input and the clean signal as the output form a pair of training samples. A large number of samples are used to train a deep learning model to fit the distribution of signal and noise data, retain the information of the clean signal, and remove the noise information to generate an optimal mapping function, so that the SISNR index between the denoised signal output by the model and the clean signal is maximized. The optimal mapping function output by the model is not the spectrogram of the denoised signal, but a mask is output, and its value is limited between [-1, 1] by the hyperbolic tangent activation function, indicating the contribution degree of the noisy signal at this time and frequency to the clean signal, which is convenient for denoising the noisy signal of different frequencies. Although it is feasible to directly estimate the spectrogram of the clean signal, better performance can be obtained by using the mask method. The network directly performs calculations using complex numbers, and the mask also uses a Complex Ratio Mask (CRM) based on complex numbers. This mask can well reconstruct the clean signal by simultaneously shrinking the divided real and imaginary components.

[0081] In one embodiment, refer to Figure 4 , Figure 4 is the specific structure of the noise reduction model described in step S20. The noisy signal \(y(t)\) collected by the sensor is used as the model input. First, the time-frequency feature spectrogram of the leakage signal to be denoised is normalized, and then it becomes \(Y(t,f)\) through the short-time Fourier transform. After a series of encoder and decoder groups for feature reconstruction of \(Y(t,f)\), a CRM mask is output. After being limited by the tanh function, the mask is multiplied pointwise with the initial \(Y(t,f)\), and the product is inverse short-time Fourier transformed to obtain the final denoised signal. After denoising, the leakage signals corresponding to No. 1, No. 2, and No. 3 are respectively denoted as

[0082] Furthermore, in order to directly process complex data, the network adopts a deep complex network module that can handle complex numbers, including a complex two-dimensional convolution module and a complex batch normalization module. Compared with traditional networks that only consider amplitude information, introducing phase information into the network has been proven to significantly enhance network performance. Taking convolution as an example, define the complex input as X = X r +iX i , given the complex convolution kernel W = A + iB with shared parameters, where X r and A represent the real part, and Xx and B represent the imaginary part. The output of the convolution operation can be defined as:

[0083] Y = (X r A - X i B) + i(X r B + X i A)

[0084] In one embodiment, Figure 4 the encoder part in Figure 5 , denoted as E, and the decoder part, denoted as D, for their specific structures refer to r , after the complex input passes through the complex two-dimensional convolution module, the complex batch normalization module, and the complex LeakyRelu module, a complex output is obtained. The difference between the encoder and the decoder is that the complex two-dimensional convolution module in the encoder module performs forward convolution operations, while the complex two-dimensional convolution module in the decoder performs inverse convolution operations. In order to use the LeakyRule activation function in the complex network, two functions are used to process the real part and the imaginary part of the complex number respectively. Let the complex input be X, and its real part and imaginary part be X i , respectively, then the complex LeakyRelu activation function CLRelu is defined as:

[0085] CLRelu(X) = LRelu(X r ) + iLRelu(X i )

[0086] In one embodiment, optionally, both the encoder and decoder modules have seven layers, and the convolution kernel size of each is 3×3. The encoder and decoder with the same number of layers are added through skip connections, that is, the real part and the imaginary part of the complex multi-dimensional feature vector are respectively subjected to addition operations. The number of input and output channels of the encoders E1 to E7 are (1, 16), (16, 32), (32, 64), (64, 128), (128, 128), (128, 256), (256, 512) respectively, and the number of input and output channels of the decoders D1 to D7 are (512, 256), (512, 128), (256, 128), (256, 64), (128, 32), (64, 16), (32, 1) respectively. The present invention does not limit the specific number of layers and parameters of the encoder and decoder, and corresponding adjustments can be made according to the data volume and the computing power of the device.

[0087] In a possible implementation manner, before obtaining the denoised first leakage signal, second leakage signal, and third leakage signal through a pre-trained denoising model for urban environment based on a complex convolution network, it further includes training the complex two-dimensional convolution module network in the encoder and decoder modules, including:

[0088] Obtain training samples, where the training samples are a number of leakage sound signals and urban noise signals;

[0089] The input of the training set of the model is the noisy leakage sound signal formed by superimposing the leakage sound signal and the urban noise signal in the time domain with a random signal-to-noise ratio of (-5, 15) dB in the time domain; [[ID=!1]]

[0090] The output of the training set of the model is the leakage sound signal corresponding to the input of the model training;

[0091] Backpropagate in the convolution network according to the independent audio source separation signal-to-noise ratio as the loss function, and update and process the weight matrix corresponding to the encoder-decoder module in combination with the gradient descent method until the loss function meets the preset conditions, and then generate the denoising network model. The larger the loss function value, the better the effect, and the specific preset value should be determined according to the sample data volume and the effect requirements.

[0092] Step S30: Perform cross-correlation calculations on the denoised first and second leakage signals and the first and third leakage signals respectively to obtain the time delay values.

[0093] It is easy to understand that for the denoised first, second, and third leakage signals The cross-correlation calculation is divided into two steps. Taking the denoised first and second leakage signals as an example, the first step is the calculation of the cross-correlation function R 12 (θ):

[0094]

[0095] Step 2: Denote the signal sampling rate as sr, change the value of θ, where the value range of θ is θ ∈ (-4 sr , 4 sr ), and the change step size is 1. Record the value of θ when the cross-correlation function R 12 (θ) reaches the maximum. After dividing it by the sampling rate sr, it is the time delay value Δt1:

[0096]

[0097] Correspondingly, obtain the time delay values of the first and third leakage signals after noise reduction processing, denoted as Δt2.

[0098] Step S40: Estimate the communication delay between the first and second sensors and the first and third sensors to obtain the time delay value after compensating for the communication delay.

[0099] It is easy to understand that the first sensor always serves as the master sensor, and the second and third sensors always serve as slave sensors. Taking the estimation of the communication delay between the first and second sensors as an example, the process is as follows: Record the local time of the first master sensor as t1, and immediately send the first handshake signal to the second slave sensor; after receiving the first handshake signal, the second slave sensor records its local time as t2; the second slave sensor records its local time as t3 and immediately sends the second handshake signal to the first master sensor; after receiving the second handshake signal, the first master sensor records its local time as t4; calculate and solve the communication delay between the first and second sensors according to the following formula, denoted as τ1:

[0100] τ1 = ((t4 - t3) + (t2 - t1)) / 2

[0101] Then (Δt1 - τ1) is the time delay value between the first and second sensors after compensating for the communication delay. Correspondingly, obtain the communication delay between the first and third sensors as τ2, and (Δt2 - τ2) is the time delay value between the first and third sensors after compensating for the communication delay.

[0102] Step S50: Obtain the distances between the first and second sensors and the second and third sensors.

[0103] It is easy to understand that taking the example of obtaining the distance between the first and second sensors, the process is as follows: Through the position acquisition module 24, obtain the longitude and latitude information of the first and second sensors; convert the longitude and latitude information into radian system representation; apply the Mercator projection method to calculate the corresponding plane coordinates of the first sensor, denoted as (x1, y1), and the corresponding plane coordinates of the second sensor, denoted as (x2, y2); calculate the Euclidean distance from the plane coordinates of the first and second sensors, which is the distance between the first and second sensors, denoted as L1:

[0104]

[0105] Correspondingly, calculate the distance between the second and third sensors, denoted as L2.

[0106] Step S60: Calculate the precise pipeline leakage position according to the obtained distance and time delay value.

[0107] It is easy to understand that according to steps S10 to S50, the time delay values (Δt1 - τ1) between the first and second sensors after compensating for communication delay and the time delay value (Δt2 - τ2) between the first and third sensors after compensating for communication delay have been obtained. The distance between the first sensor and the second sensor is L1, and the distance between the second sensor and the third sensor is L2. The distance x of the first host sensor from the pipeline leakage point, that is, the precise position of the pipeline leakage point, can be calculated through the following formula:

[0108]

[0109] Those of ordinary skill in the art can realize that the components and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but this implementation should not be considered to exceed the scope of the present invention.

[0110] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A method for parameterless positioning of urban pipeline leaks, characterized in that, The positioning method includes: Simultaneously obtaining a first leakage signal, a second leakage signal, and a third leakage signal corresponding to a first sensor, a second sensor, and a third sensor deployed on both sides of a pipeline leakage point; the first sensor, the second sensor, and the third sensor are distributed on the pipeline to be measured, the pipeline leakage point is located between the first sensor and the second sensor, the third sensor is close to the second sensor and far from the first sensor; Obtaining the noise-reduced first leakage signal, second leakage signal, and third leakage signal through a pre-trained noise reduction model for urban environments based on a complex convolutional network; Performing cross-correlation calculations on the noise-reduced first leakage signal and the second leakage signal, and the first leakage signal and the third leakage signal respectively to obtain time delay values; Performing communication delay estimation on the first sensor and the second sensor, and the first sensor and the third sensor to obtain the time delay value after compensating for communication delay; Obtaining the distances between the first sensor and the second sensor, and the second sensor and the third sensor; Calculating the exact position of the pipeline leakage point based on the obtained distance and the time delay value after compensating for communication delay.

2. The method for parameterless positioning of urban pipeline leakage according to claim 1, characterized in that, The step of obtaining the first leakage signal, second leakage signal, and third leakage signal corresponding to the first sensor, second sensor, and third sensor deployed on both sides of the pipeline leakage point while trying to ensure simultaneity includes: The first sensor is the master sensor, and the second sensor and the third sensor are slave sensors. Before the first master sensor starts to obtain the pipeline leakage point signal, it notifies the second and third slave sensors to start obtaining the pipeline leakage signal through wireless communication; The first master sensor starts to obtain the pipeline leakage point signal; After receiving the start signal sent by the first sensor, the second and third slave sensors immediately start to obtain the pipeline leakage point signal; Performing clipping processing on the obtained first leakage signal, second leakage signal, and third leakage signal respectively to ensure that the signal length is uniformly 4 seconds, discarding the part exceeding 4 seconds, and padding with zeros for the part less than 4 seconds.

3. The method for parameterless positioning of urban pipeline leakage according to claim 1, wherein, The step of obtaining the noise-reduced first leakage signal, second leakage signal, and third leakage signal through a pre-trained noise reduction model for urban environments based on a complex convolutional network includes: Performing short-time Fourier transform on the first leakage signal, the second leakage signal, and the third leakage signal respectively to obtain the time-frequency characteristic spectrogram of the leakage signal to be noise-reduced; Normalizing the time-frequency characteristic spectrogram of the leakage signal to be noise-reduced; Performing operations on the normalized time-frequency characteristic spectrogram through 7 encoder modules to obtain a compressed complex multi-dimensional feature vector; Performing operations on the compressed complex multi-dimensional feature vector through 7 decoder modules, plus the output of the corresponding encoder module with skip connection, to obtain a noise-reduced complex ratio mask; Performing a dot product operation on the noise-reduced complex ratio mask and the normalized time-frequency characteristic spectrogram to obtain a noise-reduced two-dimensional time-frequency characteristic spectrogram; Perform an inverse short-time Fourier transform on the denoised two-dimensional time-frequency feature spectrogram to obtain the denoised acoustic signal.

4. The method for parameterless positioning of urban pipeline leakage according to claim 3, wherein The encoder module and the decoder module are skip connections, including: The structures of the encoder module and the decoder module are such that the input layer is a complex number composed of the real part and the imaginary part of the feature spectrogram or a complex multi-dimensional vector feature output by the previous layer. The first layer is a complex two-dimensional convolution module, the second layer is a complex batch normalization module, the third layer is a complex LeakyRelu module, and the output layer is a complex multi-dimensional vector feature; The skip connection is an adder structure connecting the encoder and the decoder, enabling the real part and the imaginary part of the complex multi-dimensional feature vector to perform addition operations respectively.

5. The method for parameterless positioning of urban pipeline leakage according to claim 3, characterized in that Before the step of obtaining the denoised first leakage signal, second leakage signal, and third leakage signal through a pre-trained noise reduction network model based on a complex convolution network for an urban environment, it further includes training the complex two-dimensional convolution module network in the encoder and decoder modules, including: Obtain training samples, where the training samples are several leakage acoustic signals and urban noise signals; The input of the training set of the noise reduction network model is a noisy leakage acoustic signal formed by superimposing a leakage acoustic signal and an urban noise signal in the time domain with a random signal-to-noise ratio of (-5, 15) dB in the time domain; The output of the training set of the noise reduction network model is the leakage acoustic signal corresponding to the input of the model training; Automatically generate a corresponding initial weight matrix according to the initialization conditions of the encoder and decoder modules; Calculate and generate an independent audio source separation signal-to-noise ratio loss function corresponding to the encoder and decoder modules according to the training set of the noise reduction network model; Backpropagate in the convolutional network according to the loss function, and update and process the weight matrix corresponding to the encoder-decoder module in combination with the gradient descent method until the loss function meets the preset conditions, and then generate the noise reduction network model.

6. The method for parameterless positioning of urban pipeline leakage according to claim 1, wherein, The step of estimating the communication delay between the first sensor and the second sensor and between the first sensor and the third sensor includes: The first host sensor records its local time as t1 and immediately sends a first handshake signal to the slave sensor; After receiving the first handshake signal, the slave sensor records its local time as t2; The slave sensor records its local time as t3 and immediately sends a second handshake signal to the first host sensor; After receiving the second handshake signal, the host sensor records its local time as t4; Calculate and solve the communication delay τ according to the following formula: τ = ((t4 - t3) + (t2 - t1)) / 2.

7. The method for parameterless positioning of urban pipeline leakage according to claim 1, characterized in that The step of obtaining the distances between the first sensor and the second sensor and between the second sensor and the third sensor includes: Obtain the longitude and latitude information of the first sensor, the second sensor, and the third sensor; Convert the longitude and latitude information to radian measure; Apply the Mercator projection method to calculate the corresponding planar coordinates of each sensor; Calculate the Euclidean distance between the plane coordinates of the first sensor and the second sensor, and between the second sensor and the third sensor respectively.

8. The method for parameterless leakage location of urban pipelines according to claim 1, characterized in that The step of calculating the exact position of the pipeline leakage point according to the obtained distance and the time delay value includes: Define the time delay between the first leakage signal and the second leakage signal as Δt1, and the time delay between the first leakage signal and the third leakage signal as Δt2; Define the communication time delay between the first sensor and the second sensor as τ1, and the time delay between the first sensor and the third sensor as τ2; Define the distance between the first sensor and the second sensor as L1, and the distance between the second sensor and the third sensor as L2; The distance x from the first main sensor to the pipeline leakage point, that is, the exact position of the pipeline leakage point, can be calculated by the following formula:

9. An urban pipeline leakage parameter-free positioning device for implementing the method according to any one of claims 1-8, characterized in that, The urban pipeline leakage parameterless positioning device includes: A control and calculation center device, equipped with a touchable display and a wireless communication interface, running a visual user interface program, controlled by the user for the progress of the leakage positioning method, communicating with the signal acquisition device, storing the data sent by the leakage signal acquisition device, and quickly implementing the steps of the urban pipeline leakage parameterless positioning method according to any one of claims 1, 3, 4, 5, or 7; A signal acquisition device, to which the first sensor, the second sensor, and the third sensor all belong, equipped with a signal acquisition device and a storage device to execute the steps described in claim 2, equipped with a wireless communication interface and a real-time operating system, execute the steps described in claim 6, and at the same time be controlled by the control and calculation center device through wireless communication and be able to perform data transmission with it.

Images