Pipe damage identification method and device applied to water supply pipeline and electronic equipment

Abstract

Embodiments of the present disclosure disclose a pipe damage identification method, device and electronic equipment applied to a water supply pipeline. A specific implementation of the method comprises: determining an initial pipe damage region of interest according to water supply pipeline state data; collecting an infrared image and a reflected wave signal sequence corresponding to the initial pipe damage region of interest; generating a humidity thermal map according to the reflected wave signal sequence; performing pipe damage coarse positioning according to the infrared image and the humidity thermal map to obtain a target pipe damage region of interest and a target region feature map; collecting an acoustic signal array corresponding to the target pipe damage region of interest in a directional manner; performing regional correction on the target pipe damage region of interest according to the acoustic signal array and the target region feature map to obtain pipe damage description information. The implementation improves the pipe damage identification precision and accuracy.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of computer technology, and in particular to the field of damage identification, specifically to a method, apparatus, and electronic device for identifying pipe damage in water supply pipelines. Background Technology

[0002] Direct burial is one of the main methods for laying water supply pipelines. Although it does not occupy surface space, it is highly susceptible to accelerated pipe aging due to soil corrosion, leading to leaks. Currently, the common method is to locate leak points manually.

[0003] However, when using the above method, the following technical problems often arise: When a leak in a water supply pipeline is not obvious, it is difficult to directly observe the location of the leak from the surface, making it difficult to guarantee the accuracy and precision of the identification. Summary of the Invention

[0004] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0005] Some embodiments of this disclosure propose a method, apparatus, and electronic device for identifying pipe damage in water supply pipelines to address the technical problems mentioned in the background section above.

[0006] In a first aspect, some embodiments of this disclosure provide a method for identifying pipe damage in water supply pipelines. The method includes: determining an initial region of interest (ROI) for pipe damage based on water supply pipeline status data; acquiring an infrared image and a reflected wave signal sequence corresponding to the initial ROI, wherein the reflected wave signal in the reflected wave signal sequence is acquired by ground-penetrating radar; generating a humidity heat map based on the reflected wave signal sequence; performing coarse pipe damage localization based on the infrared image and the humidity heat map to obtain a target ROI and a target region feature map, wherein the target ROI is a sub-region of the initial ROI; directionally acquiring an acoustic signal array corresponding to the target ROI; and performing region correction on the target ROI based on the acoustic signal array and the target region feature map to obtain pipe damage description information, wherein the pipe damage description information includes: pipe damage location and pipe damage type.

[0007] Secondly, some embodiments of this disclosure provide a pipe damage identification device for water supply pipelines. The device includes: a determining unit configured to determine an initial region of interest (ROI) for pipe damage based on water supply pipeline status data; an acquisition unit configured to acquire an infrared image and a reflected wave signal sequence corresponding to the initial ROI, wherein the reflected wave signal in the reflected wave signal sequence is acquired by ground-penetrating radar; a generating unit configured to generate a humidity heat map based on the reflected wave signal sequence; a coarse pipe damage localization unit configured to perform coarse pipe damage localization based on the infrared image and the humidity heat map to obtain a target ROI and a target region feature map, wherein the target ROI is a sub-region of the initial ROI; a directional acquisition unit configured to directionally acquire an acoustic signal array corresponding to the target ROI; and a region correction unit configured to perform region correction on the target ROI based on the acoustic signal array and the target region feature map to obtain pipe damage description information, wherein the pipe damage description information includes: pipe damage location and pipe damage type.

[0008] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0009] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0010] The embodiments disclosed above have the following beneficial effects: the pipe loss identification method applied to water supply pipelines according to some embodiments of this disclosure improves the accuracy and precision of pipe loss identification. Specifically, firstly, an initial region of interest (ROI) for pipe loss is determined based on water supply pipeline status data. In practice, the topology of water supply pipelines is usually quite complex. In direct burial installations, manual inspection is insufficient for timely pipe loss identification. Therefore, this disclosure combines water supply pipeline status data to locate areas where pipe loss may occur, thereby quickly narrowing the identification range. Secondly, infrared images and reflected wave signal sequences corresponding to the initial ROI are acquired, where the reflected wave signals in the reflected wave signal sequence are obtained by ground-penetrating radar. In practice, even slight seepage can cause changes in humidity and temperature when pipe loss occurs. Considering the difference in signal refractive index between seepage and non-seepage areas, infrared images and reflected wave images are acquired for subsequent coarse pipe loss localization. Next, a humidity heatmap is generated based on the reflected wave signal sequence. By converting the reflected wave signals, a humidity representation in heatmap form is obtained. Furthermore, based on the aforementioned infrared image and humidity heatmap, coarse localization of pipe damage is performed to obtain the target pipe damage region of interest (ROI) and a feature map of the target region. The target ROI is a sub-region of the initial ROI. By combining the infrared image and humidity heatmap, coarse localization of pipe damage over a large area is achieved, further narrowing the identification range. In addition, an acoustic signal array corresponding to the target ROI is acquired directionally. In practice, when pipe damage occurs, seepage often generates noise such as friction noise at the leak. Therefore, acoustic signals are acquired directionally within a small area to capture pipe damage-related noise. Finally, based on the acoustic signal array and the target region feature map, the target ROI is corrected to obtain pipe damage description information, including the pipe damage location and type. By combining acoustic signals to correct the target ROI, the accuracy and precision of pipe damage location are further improved. In summary, this method improves the accuracy and precision of pipe damage identification. Attached Figure Description

[0011] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0012] Figure 1 This is a flowchart of some embodiments of the pipe loss identification method applied to water supply pipelines according to the present disclosure; Figure 2 It is a schematic diagram of the topology of the water supply pipeline. Figure 3 This is a schematic diagram illustrating the generation process of the first and second adjacency matrices corresponding to the local topology of the water supply pipeline topology diagram after binding. Figure 4 This is a schematic diagram of the basic encoder structure; Figure 5 This is a schematic diagram of the structure of an infrared image encoder; Figure 6 This is a schematic diagram of the structure of some embodiments of a pipe damage identification device applied to water supply pipelines according to the present disclosure; Figure 7 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0013] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0014] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0015] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0016] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0017] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0018] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0019] refer to Figure 1 The diagram illustrates a flow 100 of some embodiments of a pipe loss identification method for water supply pipelines according to the present disclosure. This pipe loss identification method for water supply pipelines includes the following steps: Step 101: Determine the region of interest for initial pipe loss based on the water supply pipeline status data.

[0020] In some embodiments, the execution subject (e.g., a computing device) of the pipe loss identification method applied to a water supply pipeline can determine an initial region of interest for pipe loss based on water supply pipeline status data.

[0021] Among them, the water supply pipeline status data represents the operational data of the water supply pipeline during the water supply process. Specifically, the water supply pipeline status data may include, but is not limited to, water supply pressure and water supply volume. Water supply pipeline data can be collected through metering sensors installed in the water supply pipeline. Metering sensors may include, but are not limited to, pressure sensors and flow sensors. The initial region of interest for pipeline loss is the area suspected of experiencing pipeline loss, determined by combining the water supply pipeline status data.

[0022] In practice, metering sensors can be installed at the ends of pipelines connected to the water supply line, or at locations where water supply is diverted, to collect segmented data (water supply line status data) of the water supply line. Specifically, when pipe loss occurs, the water supply pressure, water supply volume, and water flow rate will change. Therefore, by combining the water supply line status data with the data for comparison, the area where pipe loss occurs can be roughly located as the initial region of interest for pipe loss.

[0023] It should be noted that the aforementioned computing devices can be either hardware or software. When the computing device is hardware, it can be implemented as a distributed cluster consisting of multiple servers or terminal devices, or as a single server or terminal device. When the computing device is software, it can be installed within the hardware devices listed above. It can be implemented as, for example, multiple software programs or software modules used to provide distributed services, or as a single software program or software module. No specific limitations are made here. Specifically, computing devices include not only hardware or software for data processing, but also related data acquisition devices for data collection.

[0024] In some optional implementations of certain embodiments, the aforementioned execution entity determines the region of interest for initial pipe loss based on water supply pipeline status data, including: Step S1: Obtain the water supply pipeline topology diagram.

[0025] The above-mentioned water supply pipeline topology diagram represents the pipeline topology of the water supply pipeline.

[0026] In practice, the topology map of the water supply pipeline during the laying stage can be read through wired or wireless connection and used as the water supply pipeline topology map.

[0027] As an example, see Figure 2The diagram shown illustrates the topological structure of the water supply pipeline. The pipeline consists of a main supply pipeline and secondary supply pipelines. The secondary supply pipelines connect to the main supply pipeline, and their ends connect to pipeline terminals. The pipeline terminals are the points where the water supply is used. Metering sensors are installed at the locations where water supply branches off from both the main and secondary supply pipelines, as well as at the pipeline terminals.

[0028] Step S2: Clean the above water supply pipeline status data to obtain cleaned water supply pipeline status data.

[0029] Among them, the status data of the water supply pipeline after cleaning can be the status data of the water supply pipeline after data cleaning.

[0030] In practice, the above water supply pipeline status data can be cleaned by removing missing values ​​to obtain cleaned water supply pipeline status data.

[0031] Step S3: Bind the above-mentioned cleaned water supply pipeline status data with the above-mentioned water supply pipeline topology map to obtain the bound water supply pipeline topology map.

[0032] In practice, since the status data of the water supply pipeline is obtained by the metering sensor, the status data of the cleaned water supply pipeline can be bound to the corresponding metering sensor to obtain the topology map of the water supply pipeline after binding.

[0033] Step S4: Extract the topology feature set corresponding to the topology map of the water supply pipeline after binding.

[0034] In this set of topological graph features, each feature corresponds to a different receptive field. The topological graph features consist of pipeline topological features, pipeline pressure features, and pipeline flow characteristics. Pipeline topological features characterize the topological structure of the water supply pipeline. Pipeline pressure features characterize the water supply pressure characteristics of the water supply pipeline. Pipeline flow characteristics characterize the water supply flow characteristics of the water supply pipeline.

[0035] In practice, the topology map of the bound water supply pipeline can be stored in the form of an adjacency matrix. Assuming the matrix size is N×N, where N is the number of metering sensors, two adjacency matrices (a first adjacency matrix and a second adjacency matrix) can be constructed. The first adjacency matrix represents the topological structure of the bound water supply pipeline, and the second adjacency matrix represents the post-cleaning water supply pipeline status data. Specifically, a first feature pyramid network can be used to perform multi-scale feature extraction on the first adjacency matrix to obtain the pipeline topology features included in each topology map feature set. Similarly, a second feature pyramid network can be used to perform multi-scale feature extraction on the second adjacency matrix to obtain the pipeline pressure and flow rate features included in each topology map feature set. The second feature pyramid network consists of two sub-feature pyramid networks. Due to the correlation between pipeline pressure and flow rate, the two sub-feature pyramid networks share parameters to perform multi-scale feature extraction on the pipeline pressure and flow rate dimensions respectively, based on the second adjacency matrix. The network structures of the two sub-feature pyramid networks are identical. The sub-feature pyramid network has the same network structure as the first feature pyramid network. Specifically, the first feature pyramid network consists of 7 downsampling layers, where the even-numbered downsampling layers are used for feature output.

[0036] As an example, see Figure 3 The diagram shows the generation process of the first and second adjacency matrices corresponding to the local topology of the bound water supply pipeline topology. Water supply pipelines P2 and P3 are connected in parallel to water supply pipeline P1. A metering sensor S1 is installed at the end of water supply pipeline P1, a metering sensor S2 is installed at the end of water supply pipeline P2, and a metering sensor S3 is installed at the end of water supply pipeline P3. Metering sensor P1 is bound to state data R1, which is included in the water supply pipeline state data. Metering sensor S2 is bound to state data R2, and metering sensor S3 is bound to state data R3, which is included in the water supply pipeline state data. Therefore, state data R1 can characterize the water supply status of water supply pipeline P1, state data R2 can characterize the water supply status of water supply pipeline P2, and state data R3 can characterize the water supply status of water supply pipeline P3. Figure 2 The local topology shown has two 3×3 adjacency matrices: the first adjacency matrix and the second adjacency matrix. For the first adjacency matrix, the value is 1 if there is a unidirectional connection between two water supply pipes, and 0 otherwise. For the second adjacency matrix, the value is the state data if the water supply pipe has corresponding state data, and 0 otherwise. Figure 3For example, the first adjacency matrix can be [[0,1,1],[0,0,0],[0,0,0]]. The second adjacency matrix can be [[R1,0,0],[0,R2,0],[0,0,R3]]. Here, "R1", "R2", and "R3" in the second adjacency matrix represent state data R1, state data R2, and state data R3, respectively.

[0037] Step S5: Locate the region of interest for pipe loss based on the feature set of the topology map above to obtain the initial region of interest for pipe loss.

[0038] In practice, three anchor-frame-based position regressors can be used to locate the region of interest (ROI) for pipe loss. Specifically, since the topology map features, including pipeline topology, pressure, and flow characteristics, are output by downsampling layers in the even-numbered layers of the first and second feature pyramid networks, three anchor-frame-based position regressors are used for position regression. Specifically, before the pipeline topology, pressure, and flow characteristics are input into the corresponding anchor-frame-based position regressors, feature overlay is required. Since the feature scales are consistent, a weighted overlay method can be used.

[0039] Step 102: Acquire the infrared image and reflected wave signal sequence corresponding to the region of interest of the initial pipe loss.

[0040] In some embodiments, the aforementioned execution entity may acquire infrared images and reflected wave signal sequences corresponding to the region of interest in the initial pipe loss.

[0041] The infrared images are captured by an infrared camera. The reflected wave signals in the reflected wave signal sequence are obtained by ground-penetrating radar.

[0042] In practice, infrared images can be acquired by pointing an infrared camera towards the region of interest (ROI) of initial pipe damage, and ground-penetrating radar (GPR) can be used to acquire zoned reflected wave signals towards the ROI. In particular, the infrared camera and GPR acquire data synchronously to ensure data alignment. For example, a drone or autonomous robot equipped with both an infrared camera and GPR can be used to acquire infrared images and reflected wave signals from the ROI of initial pipe damage.

[0043] Step 103: Generate a humidity heatmap based on the reflected wave signal sequence.

[0044] In some embodiments, the aforementioned execution entity can generate a humidity heatmap based on a sequence of reflected wave signals.

[0045] Among them, the humidity heatmap represents the humidity difference corresponding to the region of interest with initial pipe loss in the form of a heatmap.

[0046] In practice, the dielectric constant of water is much greater than that of dry soil and air. Specifically, the dielectric constant of water is typically around 80, that of air is typically 1, and that of dry soil is typically between 4 and 9. Therefore, by combining reflected wave signals and mapping them to the corresponding dielectric constants, the humidity differences within a region can be depicted. In particular, a fixed-distance method can be used to collect reflected wave signal sequences via ground-penetrating radar, and the dielectric constant can be estimated by combining the reflected wave signals. This can then be converted into a humidity heatmap to obtain a humidity heatmap.

[0047] In some optional implementations of some embodiments, the execution entity generates a humidity heatmap based on the reflected wave signal sequence, including: Step S1: Based on the above reflected wave signal sequence, determine the mean reflected wave signal and background signal noise.

[0048] The mean reflected wave signal characterizes the overall reflection characteristics of the reflected wave signal sequence. Background signal noise characterizes the environmental noise present during the acquisition of the reflected wave signal sequence. Specifically, since the reflected wave signal sequence is acquired using ground-penetrating radar at a fixed distance, environmental noise is inevitably collected. Therefore, it is necessary to remove the environmental noise to improve the accuracy of subsequent dielectric constant estimation.

[0049] In practice, firstly, considering the slope constraints of water supply pipelines, the corresponding surface soil cover often does not exhibit significant undulations. Therefore, to improve calculation speed, an arithmetic mean calculation method is used, combining the reflected wave signal sequence to calculate the mean reflected wave signal. Secondly, environmental noise is mainly generated during propagation between air and the ground surface. Therefore, a time window for propagation between air and the ground surface can be constructed, and each reflected wave signal in the reflected wave signal sequence can be extracted. The arithmetic mean of the multiple extracted sub-reflected wave signals (related to environmental noise) within the time window is then used as the background environmental noise.

[0050] Step S2: For each reflected wave signal in the above reflected wave signal sequence, perform the following processing steps: Step S21: Based on the above mean reflected wave signal, perform drift elimination on the above reflected wave signal to obtain the drift-eliminated reflected wave signal.

[0051] Wherein, the reflected wave signal after drift cancellation = reflected wave signal - mean reflected wave signal. This method can effectively suppress zero-mean random noise and improve the signal-to-noise ratio by a factor of √N.

[0052] Step S22: Based on the background signal noise, perform noise removal on the drift-eliminated reflected wave signal to obtain the first noise-removed reflected wave signal.

[0053] Wherein, the reflected wave signal after first noise removal = reflected wave signal after drift cancellation - background signal noise × noise weight. The noise weight ranges from (0,1). For example, the noise weight can be 0.95.

[0054] Step S23: Perform a mixed filtering on the first noise-removed reflected wave signal to obtain the second noise-removed reflected wave signal.

[0055] In practice, the reflected wave signal after the first noise removal can be mixed and filtered using bandpass filtering and median filtering to obtain the reflected wave signal after the second noise removal. Specifically, bandpass filtering can effectively filter low-frequency and high-frequency noise, while median filtering can effectively filter random noise.

[0056] Step S24: Extract water content features from the reflected wave signal after the second noise removal to obtain water content features.

[0057] Among them, the water content characteristic represents the soil water content of the signal acquisition area corresponding to the reflected wave signal after the second noise removal.

[0058] In practice, water content characteristics can be generated by combining the dielectric constant estimation of the reflected wave signal after the second noise removal.

[0059] Step S3: Perform regional projection based on the obtained moisture content characteristic sequence to obtain the initial humidity thermogram.

[0060] The initial humidity heatmap and the humidity heatmap have the same size.

[0061] In practice, since the reflected wave signal sequence is collected in zones, the water content characteristics (dielectric constant) corresponding to the reflected wave signal can be mapped to the center of the corresponding region to obtain the initial humidity thermogram.

[0062] Step S4: Perform pseudo-color mapping on the initial humidity heatmap to obtain a humidity heatmap.

[0063] In practice, the initial humidity heatmap is first continuously interpolated, and then the dielectric constant within the initial humidity heatmap is mapped to color values ​​to obtain the humidity heatmap. Specifically, the larger the dielectric constant, the more bluish the corresponding color value; the smaller the dielectric constant, the more reddish the corresponding color value.

[0064] Step 104: Perform coarse localization of pipe damage based on infrared images and humidity thermal maps to obtain the target pipe damage region of interest and the target region feature map.

[0065] In some embodiments, the aforementioned execution entity can perform coarse pipe damage localization based on infrared images and humidity thermal maps to obtain the target pipe damage region of interest and the target region feature map.

[0066] The target region of interest (ROI) is a sub-region of the initial ROI. Specifically, the target ROI is a narrowed area within the initial ROI that is suspected of having pipe damage. The target region feature map is the feature map corresponding to the target ROI.

[0067] In practice, the SM3Det model can be used to coarsely locate pipe damage using infrared images and humidity heat maps as inputs, obtain the target pipe damage region of interest, and use the feature map corresponding to the target pipe damage region of interest as the target region feature map.

[0068] In some optional implementations of certain embodiments, the aforementioned execution entity performs coarse pipe damage localization based on infrared images and humidity thermal maps to obtain the target pipe damage region of interest and a target region feature map, including: Step S1: The infrared image is encoded using an infrared image encoder to obtain infrared image features.

[0069] Among them, infrared image features are the image features corresponding to the extracted infrared images.

[0070] Step S2: The humidity heatmap is image encoded using a humidity heatmap encoder to obtain humidity heatmap features.

[0071] The infrared image encoder and the humidity heat map encoder have the same model structure. The infrared image encoder consists of 7 sub-encoders, which are composed of a basic encoder and a spatial attention encoder.

[0072] As an example, see Figure 4 The diagram shows the structure of the basic encoder, which consists of: a depthwise separable convolutional layer, a normalization layer, an activation function, a global average pooling layer, a normalization layer, a 2D convolutional layer, a 2D convolutional layer, a global average pooling layer, a normalization layer, and a 2D convolutional layer. The spatial attention encoder uses a Spatial Attention module. The sub-encoders, including the basic encoder and the spatial attention encoder, are connected in series. See also... Figure 5The schematic diagram of the infrared image encoder shown illustrates that it consists of sub-encoders E1, E2, E3, E4, E5, E6, and E7. Sub-encoder E1 comprises a base encoder B1 and a spatial attention encoder S1. Sub-encoder E2 comprises a base encoder B2 and a spatial attention encoder S2. Sub-encoder E3 comprises a base encoder B3 and a spatial attention encoder S3. Sub-encoder E4 comprises a base encoder B4 and a spatial attention encoder S4. Sub-encoder E5 comprises a base encoder B5 and a spatial attention encoder S5. Sub-encoder E6 comprises a base encoder B6 and a spatial attention encoder S6. Sub-encoder E7 comprises a base encoder B7 and a spatial attention encoder S7. The seven sub-encoders are connected serially. A skip connection is provided between spatial attention encoders S1 and S3. A skip connection is provided between spatial attention encoders S3 and S5. A skip connection is provided between spatial attention encoders S5 and S7. A skip connection is established between spatial attention encoder S2 and spatial encoder S4. A skip connection is also established between spatial attention encoder S4 and spatial encoder S6. Since the model structures of the infrared image encoder and the humidity heatmap encoder are consistent, the model structure of the humidity heatmap encoder will not be described again. Specifically, since infrared images and humidity heatmaps are two different modalities of input data, the infrared image encoder and the humidity heatmap encoder do not use a shared weight approach, but encode independently. Furthermore, designing infrared image encoders and humidity heatmap encoders with consistent model structures can reduce the training and update costs of the model and ensure that the extracted infrared image features and humidity heatmap features are in the same feature space and maintain the same feature dimension. Further, the decoupling design of sub-encoders, basic encoders, and spatial attention encoders facilitates model structure updates according to actual needs. Simultaneously, the spatial attention encoder is introduced to increase attention to abnormal regions. Finally, skip connections prevent feature forgetting.

[0073] Step S3: Use a cross-modal attention fusion processor to fuse the infrared image features and the humidity heatmap features to obtain fused image features.

[0074] Among them, the fused image features are the image features obtained by fusing infrared image features and the humidity thermogram features mentioned above.

[0075] In practice, the cross-modal attention fusion engine, based on an attention mechanism, calculates the feature weights corresponding to infrared image features and humidity heatmap features, and then uses a weighted feature fusion method to obtain the fused image features. Specifically, firstly, through linear projection, the infrared image features are converted into a Query vector (Q), and the humidity heatmap features are converted into a Key vector (K) and a Value vector (V). The linear projection process involves three learnable linear layers. Next, attention weights are calculated using the Query and Key vectors. Finally, the Value vector is weighted using these attention weights to obtain the fused image features.

[0076] Step S4: Generate the initial region of interest for tube loss based on the image decoder and the fused image features described above.

[0077] The aforementioned infrared image encoder, humidity heatmap encoder, cross-modal attention fusion unit, and image decoder are included in the coarse pipe loss localization model. Specifically, the image decoder is an upsampling network based on a feature pyramid network structure, and the initial region of interest for pipe loss is determined using a cross-entropy loss function. The infrared image encoder, humidity heatmap encoder, cross-modal attention fusion unit, and image decoder, as a whole (the coarse pipe loss localization model), are trained in a supervised manner.

[0078] Step S5: Extract the local image features from the above fused image features that correspond to the initial region of interest for pipe loss, and use them as the target region feature map.

[0079] In practice, local image features corresponding to the location of the initial region of interest in pipe loss can be cropped from the fused image and used as the feature map of the target region.

[0080] Step 105: Acquire the acoustic signal array corresponding to the region of interest of the target pipe loss.

[0081] In some embodiments, the aforementioned execution entity can directionally acquire acoustic signal arrays corresponding to the region of interest of the target pipe loss.

[0082] Among them, the acoustic signal is the sound signal collected towards the region of interest of the target pipe damage.

[0083] In practice, acoustic signals can be collected directionally from the region of interest (ROI) of the target pipe loss using a pickup sensor, resulting in an acoustic signal array. Since the ROI of the target pipe loss is significantly smaller than the initial ROI, the area of ​​the directional acoustic signal array can be reduced.

[0084] In some optional implementations of some embodiments, the above-mentioned execution entity directionally acquires the acoustic signal array corresponding to the region of interest of the target pipe loss, including: Step S1: Divide the region of interest of the target pipe loss into regions to obtain the acquisition region array.

[0085] In this context, the acquisition areas in the aforementioned acquisition area array all have the same size. For example, assuming the area size of the region of interest for the target pipe loss is W×H, the area size of the acquisition area array can be w×h.

[0086] In practice, a sliding window can be constructed with w as the horizontal movement step size and h as the vertical movement step size to divide the region of interest of the target pipe loss into regions, thus obtaining a data acquisition area array.

[0087] Step S2: For each acquisition area in the above acquisition area array, perform the following acquisition steps: Step S21: Acquire environmental noise signals for the acquisition area using the surround acoustic sensor included in the acoustic signal acquisition device.

[0088] The acoustic signal acquisition equipment is used to acquire acoustic signal arrays. It comprises array-type acoustic sensors and surround-type acoustic sensors. The array-type acoustic sensors are oriented perpendicularly to the acquisition area. The surround-type acoustic sensors are oriented perpendicularly to the array-type sensors. Specifically, the array-type acoustic sensors consist of 3×3 acoustic sensors. The surround-type acoustic sensors consist of four acoustic sensors. In the surround-type acoustic sensor, the angle between any two sensors is 90 degrees. The acoustic sensors utilize directional microphones.

[0089] In practice, the ambient noise within a 360-degree range centered on the acquisition area can be collected using the four acoustic sensors in a surround acoustic sensor as an ambient noise signal.

[0090] Step S22: Generate an adaptive signal gain based on the above environmental noise signal.

[0091] The adaptive signal gain is used to control the sensitivity of the array-type acoustic sensor. The adaptive signal gain is inversely proportional to the noise intensity represented by the ambient noise signal; that is, the higher the noise intensity represented by the ambient noise signal, the lower the adaptive signal gain; conversely, the lower the noise intensity represented by the ambient noise signal, the higher the adaptive signal gain. This ensures low sensitivity under high noise intensity and high sensitivity under low noise intensity. Specifically, the adaptive signal gain = maximum gain threshold / (1 + α × ambient noise signal), where the maximum gain threshold is a preset, allowable maximum gain, and α is a control coefficient that controls the degree of inverse proportionality.

[0092] Step S23: Based on the adaptive signal gain, control the array acoustic sensor included in the acoustic signal acquisition device to acquire an initial acoustic signal set for the acquisition area.

[0093] In practice, the adaptive signal gain can be used as the analog gain to control the acquisition of the initial acoustic signal set for the acquisition area of ​​the array acoustic sensor, including the acoustic sensor.

[0094] Step S24: Perform signal preprocessing on each initial acoustic signal in the above initial acoustic signal group to generate a preprocessed acoustic signal, thus obtaining a preprocessed acoustic signal group.

[0095] In practice, signal alignment can be used to preprocess each initial acoustic signal in the initial acoustic signal group to generate a preprocessed acoustic signal, thus obtaining a preprocessed acoustic signal group. This ensures that subsequent signal quality evaluations are within the same signal acquisition window, thereby guaranteeing evaluation consistency.

[0096] Step S25: Generate signal quality evaluation information for each preprocessed acoustic signal in the above preprocessed acoustic signal group.

[0097] Among them, signal quality evaluation information characterizes the signal quality of the preprocessed acoustic signal.

[0098] In practice, an RNN (Recurrent Neural Network) model can be trained to generate corresponding average signal quality information from preprocessed acoustic signals. The RNN model is connected to a multi-classifier to output signal quality evaluation information. The signal quality evaluation information corresponds to signal quality values ​​ranging from [0,1]. Specifically, there are 10 categories: 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, and 1. The RNN model and multi-classifier are trained using supervised training, thus enabling rapid signal quality classification of preprocessed acoustic signals through a lightweight network structure.

[0099] Step S26: Based on the signal quality evaluation information corresponding to the preprocessed acoustic signal, perform signal fusion on each preprocessed acoustic signal in the above preprocessed acoustic signal group to obtain the acoustic signal in the acoustic signal array that corresponds to the above acquisition area.

[0100] One approach is to use a weighted fusion method, with signal quality evaluation information as the weighting factor, to average and weight each preprocessed acoustic signal in the preprocessed acoustic signal group, thereby obtaining the acoustic signal in the acoustic signal array that corresponds to the aforementioned acquisition area.

[0101] Step 106: Based on the acoustic signal array and the target area feature map, perform region correction on the target pipe loss region of interest to obtain pipe loss description information.

[0102] In some embodiments, the aforementioned execution entity can perform region correction on the target pipe loss region of interest based on the acoustic signal array and the target region feature map to obtain pipe loss description information.

[0103] The pipe loss description information indicates the location of the water supply pipeline with pipe loss and the corresponding type of pipe loss. This pipe loss description information includes: pipe loss location and pipe loss type. The pipe loss location indicates the location of the water supply pipeline with pipe loss.

[0104] In practice, since acoustic signal arrays are acquired in partitioned manner, they can be converted into feature maps and overlaid with feature maps of the target region. Then, target detection can be performed using a model such as YOLO (You Only Look Once) to obtain new regions of interest and pipe damage types, serving as pipe damage description information. This local overlay method enhances the feature representation of suspected pipe damage areas, thereby improving the localization accuracy of these areas. Furthermore, converting acoustic signals into feature maps facilitates feature fusion between multimodal data.

[0105] In some optional implementations of certain embodiments, the execution entity performs region correction on the target pipe loss region of interest based on the acoustic signal array and the target region feature map to obtain pipe loss description information, including: Step S1: Encode the acoustic signals in the acoustic signal array to obtain an acoustic signal feature map.

[0106] Among them, the acoustic signal feature map is the feature map corresponding to the acoustic signal.

[0107] In practice, because acoustic signals are acquired in partitioned areas within an acoustic signal array, the acoustic signal can be represented as a feature representation of a two-dimensional region. Specifically, features can be extracted from the acoustic signal using time-domain feature extraction, and the multiple time-domain features obtained are then stitched together in two dimensions according to the corresponding partitioned regions to obtain an acoustic signal feature map.

[0108] Step S2: Based on the above-mentioned fused image features, perform feature map sampling on the above-mentioned acoustic signal feature map to obtain the sampled acoustic signal feature map.

[0109] The feature map size of the fused image features is the same as the feature map size of the sampled acoustic signal feature map.

[0110] In practice, because the feature map size of the acoustic signal feature map differs from that of the feature map of the fused image, it is necessary to upsample the acoustic signal feature map to ensure scale consistency in subsequent feature fusion. This upsampling process yields a sampled acoustic signal feature map. Simultaneously, the feature map upsampling process also enables deep feature extraction from the feature representation of the acoustic signal feature map.

[0111] Step S3: Perform local feature fusion based on the acoustic signal feature map and the target region feature map to obtain multimodal fusion features.

[0112] Since the acoustic signal feature map and the fused image feature map correspond to the same feature scale expression, the target region feature map can be superimposed on the acoustic signal feature map according to the feature map position of the target region feature map in the fused image feature map, as a multimodal fusion feature. In this way, the feature expression of the suspected pipe damage area can be enhanced.

[0113] Step S4: Perform fine-grained pipe loss localization based on the above multimodal fusion features to obtain regional offset information; Among them, the regional offset information represents the pipe loss location obtained based on the multimodal fusion features.

[0114] In practice, a location regressor can be used to obtain region offset information by taking multimodal fusion features as input. This location regressor can be constructed based on the cross-entropy loss function.

[0115] Step S5: Based on the above-mentioned regional offset information, update the region of interest for the target pipe loss to obtain pipe loss description information.

[0116] In practice, the region of interest of the target pipe loss can be updated based on the region offset information, and this updated region location can be used as the pipe loss location included in the pipe loss description information. The local features corresponding to the pipe loss location in the multimodal fusion features are then used as the input to a multi-classifier to generate the pipe loss type included in the pipe loss description information. The upsampling network used in step S2, the location regressor in step S4, and the multi-classifier in step S5 are trained as a whole using supervised training.

[0117] In some optional implementations of some embodiments, the above method further includes: Step S1: Obtain the pipeline description information of the local water supply pipeline corresponding to the above-mentioned pipeline loss location.

[0118] In practice, the pipeline description information of the local water supply pipeline corresponding to the above-mentioned pipe loss location can be obtained through wired or wireless connections.

[0119] The pipeline description information includes: pipeline burial depth, pipeline specifications, and pipeline importance. Pipeline specifications characterize the specifications of the local water supply pipeline corresponding to the location of pipeline loss; specifically, these specifications may include parameters such as pipe type, pipe diameter, and pipe flow rate. Pipeline importance characterizes the significance of the local water supply pipeline corresponding to the location of pipeline loss. Pipeline importance can be preset manually.

[0120] Step S2: Generate updated pipe loss locations based on the above-mentioned pipe loss locations and the above-mentioned pipe burial depths.

[0121] In practice, since the location of pipe damage is a local water supply pipeline determined with the ground as the horizontal plane, and lacks depth information, the location of pipe damage is transformed into a three-dimensional location expression by combining the pipeline burial depth.

[0122] Step S3: Generate a pipeline loss handling priority based on the above-mentioned pipeline loss type and pipeline importance.

[0123] Among them, the priority of pipe damage treatment indicates the priority of treatment for the local water supply pipeline corresponding to the location of pipe damage. The higher the priority of pipe damage treatment, the higher the urgency of treating the local water supply pipeline corresponding to the location of pipe damage.

[0124] In practice, a priority decision tree can be constructed, using the type of pipeline loss and the importance of the pipeline as inputs, to generate priority for pipeline loss handling.

[0125] Step S4: Generate processing guidance information based on the updated pipe loss location and the above pipe specification information.

[0126] The aforementioned guidance information includes pipeline control guidance information and pipeline repair guidance information. Pipeline control guidance information is used to control other related water supply pipelines, centered on the local water supply pipeline corresponding to the location of the pipeline damage, including but not limited to operations such as pipeline flow restriction and throttling. Pipeline repair guidance information guides repair personnel in excavating at the location of the repaired pipeline damage.

[0127] In practice, pre-set rule triggers, combined with updated pipe loss location information and pipeline specification information, can be used to trigger processing guidance information.

[0128] Step S5: Generate a pipeline loss handling task based on the above pipeline loss handling priority and the above handling guidance information.

[0129] Among them, the pipeline damage handling task is the repair task to be assigned to the repair personnel. In practice, the pipeline damage handling priority and handling guidance information can be encapsulated into a task, which is then added to the task queue for task distribution to the repair personnel.

[0130] The embodiments disclosed above have the following beneficial effects: the pipe loss identification method applied to water supply pipelines according to some embodiments of this disclosure improves the accuracy and precision of pipe loss identification. Specifically, firstly, an initial region of interest (ROI) for pipe loss is determined based on water supply pipeline status data. In practice, the topology of water supply pipelines is usually quite complex. In direct burial installations, manual inspection is insufficient for timely pipe loss identification. Therefore, this disclosure combines water supply pipeline status data to locate areas where pipe loss may occur, thereby quickly narrowing the identification range. Secondly, infrared images and reflected wave signal sequences corresponding to the initial ROI are acquired, where the reflected wave signals in the reflected wave signal sequence are obtained by ground-penetrating radar. In practice, even slight seepage can cause changes in humidity and temperature when pipe loss occurs. Considering the difference in signal refractive index between seepage and non-seepage areas, infrared images and reflected wave images are acquired for subsequent coarse pipe loss localization. Next, a humidity heatmap is generated based on the reflected wave signal sequence. By converting the reflected wave signals, a humidity representation in heatmap form is obtained. Furthermore, based on the aforementioned infrared image and humidity heatmap, coarse localization of pipe damage is performed to obtain the target pipe damage region of interest (ROI) and a feature map of the target region. The target ROI is a sub-region of the initial ROI. By combining the infrared image and humidity heatmap, coarse localization of pipe damage over a large area is achieved, further narrowing the identification range. In addition, an acoustic signal array corresponding to the target ROI is acquired directionally. In practice, when pipe damage occurs, seepage often generates noise such as friction noise at the leak. Therefore, acoustic signals are acquired directionally within a small area to capture pipe damage-related noise. Finally, based on the acoustic signal array and the target region feature map, the target ROI is corrected to obtain pipe damage description information, including the pipe damage location and type. By combining acoustic signals to correct the target ROI, the accuracy and precision of pipe damage location are further improved. In summary, this method improves the accuracy and precision of pipe damage identification.

[0131] Further reference Figure 6 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a pipe loss identification device applied to water supply pipelines. These device embodiments are similar to... Figure 1 Corresponding to the method embodiments shown, the pipe damage identification device applied to water supply pipelines can be specifically applied to various electronic devices.

[0132] like Figure 6As shown, a pipe damage identification device 600 for water supply pipelines in some embodiments includes: a determining unit 601, a acquiring unit 602, a generating unit 603, a coarse pipe damage localization unit 604, a directional acquiring unit 605, and a region correction unit 606. The determining unit 601 is configured to determine an initial region of interest (ROI) for pipe damage based on water supply pipeline status data. The acquiring unit 602 is configured to acquire an infrared image and a reflected wave signal sequence corresponding to the initial ROI, wherein the reflected wave signal in the reflected wave signal sequence is acquired by ground-penetrating radar. The generating unit 603 is configured to generate a humidity sensor based on the reflected wave signal sequence. The system includes a heat map; a pipe damage coarse localization unit 604, configured to perform coarse pipe damage localization based on the infrared image and the humidity heat map, obtaining a target pipe damage region of interest and a target region feature map, wherein the target pipe damage region of interest is a sub-region of the initial pipe damage region of interest; a directional acquisition unit 605, configured to directionally acquire the acoustic signal array corresponding to the target pipe damage region of interest; and a region correction unit 606, configured to perform region correction on the target pipe damage region of interest based on the acoustic signal array and the target region feature map, obtaining pipe damage description information, wherein the pipe damage description information includes: pipe damage location and pipe damage type.

[0133] It is understandable that the units described in the pipe damage identification device 600 applied to water supply pipelines are similar to those in the reference device. Figure 1 The steps in the described method correspond to each other. Therefore, the operations, features, and beneficial effects described above for the method are also applicable to the pipe loss identification device 600 and the units contained therein applied to water supply pipelines, and will not be repeated here.

[0134] The following is for reference. Figure 7 It shows a schematic diagram of the structure of an electronic device (e.g., a computing device) 700 suitable for implementing some embodiments of the present disclosure. Figure 7 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0135] like Figure 7 As shown, the electronic device 700 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 701, which can perform various appropriate actions and processes according to a program stored in a read-only memory 702 or a program loaded from a storage device 708 into a random access memory 703. The random access memory 703 also stores various programs and data required for the operation of the electronic device 700. The processing unit 701, the read-only memory 702, and the random access memory 703 are interconnected via a bus 704. An input / output interface 705 is also connected to the bus 704.

[0136] Typically, the following devices can be connected to the input / output interface 705: input devices 706 including, for example, a touchscreen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; output devices 707 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 708 including, for example, magnetic tape, hard disk, etc.; and communication devices 709. Communication device 709 allows electronic device 700 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 7 An electronic device 700 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 7 Each box shown can represent a device or multiple devices as needed.

[0137] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 709, or installed from a storage device 708, or installed from a read-only memory 702. When the computer program is executed by the processing device 701, it performs the functions defined in the methods of some embodiments of this disclosure.

[0138] It should be noted that, in some embodiments of this disclosure, the computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0139] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0140] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to: determine an initial region of interest (ROI) for pipe damage based on water supply pipeline status data; acquire an infrared image and a reflected wave signal sequence corresponding to the initial ROI, wherein the reflected wave signal in the reflected wave signal sequence is acquired by ground-penetrating radar; generate a humidity heat map based on the reflected wave signal sequence; perform coarse pipe damage localization based on the infrared image and the humidity heat map to obtain a target ROI and a target region feature map, wherein the target ROI is a sub-region of the initial ROI; directionally acquire an acoustic signal array corresponding to the target ROI; and perform region correction on the target ROI based on the acoustic signal array and the target region feature map to obtain pipe damage description information, wherein the pipe damage description information includes: pipe damage location and pipe damage type.

[0141] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and Python, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0142] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0143] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0144] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for identifying pipe loss in water supply pipelines, characterized in that, include: Based on the water supply pipeline status data, determine the region of interest for initial pipeline loss; Infrared images and reflected wave signal sequences corresponding to the region of interest of the initial pipe loss are acquired, wherein the reflected wave signals in the reflected wave signal sequence are acquired by ground penetrating radar; A humidity heatmap is generated based on the reflected wave signal sequence; Based on the infrared image and the humidity heat map, coarse pipe damage localization is performed to obtain the target pipe damage region of interest and the target region feature map, wherein the target pipe damage region of interest is a sub-region of the initial pipe damage region of interest; Acoustic signal arrays corresponding to the region of interest of the target pipe loss are acquired in a directional manner; Based on the acoustic signal array and the target region feature map, region correction is performed on the target pipe loss region of interest to obtain pipe loss description information, wherein the pipe loss description information includes: pipe loss location and pipe loss type.

2. The pipe loss identification method for water supply pipelines according to claim 1, characterized in that, The method further includes: Obtain the pipeline description information of the local water supply pipeline corresponding to the location of the pipe loss, wherein the pipeline description information includes: pipeline burial depth, pipeline specification information and pipeline importance; Based on the location of the pipe loss and the burial depth of the pipeline, an updated location of the pipe loss is generated; Based on the type of pipe loss and the importance of the pipeline, a pipe loss handling priority is generated; Based on the updated location of pipe damage and the pipe specification information, processing guidance information is generated, wherein the processing guidance information includes: pipe control guidance information and pipe repair guidance information; Based on the pipeline loss handling priority and the handling guidance information, a pipeline loss handling task is generated.

3. The pipe loss identification method for water supply pipelines according to claim 2, characterized in that, The step of determining the initial region of interest for pipe loss based on water supply pipeline status data includes: Obtain a water supply pipeline topology diagram, wherein the water supply pipeline topology diagram represents the pipeline topology structure of the water supply pipeline; The water supply pipeline status data is cleaned to obtain cleaned water supply pipeline status data. The status data of the cleaned water supply pipeline is bound to the topology map of the water supply pipeline to obtain the bound topology map of the water supply pipeline; Extract the topology feature set corresponding to the topology map of the bound water supply pipeline. Each topology feature in the topology feature set corresponds to a different receptive field. The topology feature consists of pipeline topology features, pipeline pressure features, and pipeline flow features. Based on the feature set of the topology map, the region of interest for pipe loss is located to obtain the initial region of interest for pipe loss.

4. The pipe loss identification method for water supply pipelines according to claim 3, characterized in that, The step of generating a humidity heatmap based on the reflected wave signal sequence includes: Based on the reflected wave signal sequence, the mean reflected wave signal and background signal noise are determined; For each reflected wave signal in the reflected wave signal sequence, the following processing steps are performed: Based on the mean reflected wave signal, the reflected wave signal is drift-eliminated to obtain the drift-eliminated reflected wave signal; Based on the background signal noise, noise removal is performed on the drift-eliminated reflected wave signal to obtain a first noise-removed reflected wave signal. The first noise-removed reflected wave signal is mixed and filtered to obtain the second noise-removed reflected wave signal; The water content features are extracted from the reflected wave signal after the second noise removal; Based on the obtained moisture content characteristic sequence, regional projection is performed to obtain an initial humidity thermogram; The initial humidity heatmap is subjected to pseudo-color mapping to obtain a humidity heatmap.

5. The pipe loss identification method for water supply pipelines according to claim 4, characterized in that, The step of performing coarse pipe damage localization based on the infrared image and the humidity thermogram to obtain the target pipe damage region of interest and the target region feature map includes: The infrared image is encoded using an infrared image encoder to obtain infrared image features; The humidity heatmap is image encoded by a humidity heatmap encoder to obtain humidity heatmap features. The infrared image encoder and the humidity heatmap encoder have the same model structure. The infrared image encoder consists of 7 sub-encoders, which are composed of a basic encoder and a spatial attention encoder. The infrared image features and the humidity thermogram features are fused using a cross-modal attention fusion processor to obtain fused image features; An initial region of interest for pipe loss is generated based on the image decoder and the fused image features, wherein the infrared image encoder, the humidity heat map encoder, the cross-modal attention fusion unit, and the image decoder are included in the coarse pipe loss localization model; Extract the local image features from the fused image features that correspond to the initial region of interest for pipe loss, and use them as the target region feature map.

6. The pipe loss identification method for water supply pipelines according to claim 5, characterized in that, The acoustic signal array for directional acquisition of the region of interest corresponding to the target pipe loss includes: The target pipe loss region of interest is divided into regions to obtain a data acquisition region array, wherein the data acquisition regions in the data acquisition region array have the same region size; For each acquisition area in the acquisition area array, perform the following acquisition steps: The ambient noise signal for the acquisition area is acquired using the surround acoustic sensor included in the acoustic signal acquisition device. Based on the environmental noise signal, an adaptive signal gain is generated; Based on the adaptive signal gain, the array of acoustic sensors included in the acoustic signal acquisition device is controlled to acquire an initial set of acoustic signals for the acquisition area; Each initial acoustic signal in the initial acoustic signal group is preprocessed to generate a preprocessed acoustic signal, thus obtaining the preprocessed acoustic signal group; Generate signal quality evaluation information for each preprocessed acoustic signal in the preprocessed acoustic signal group; Based on the signal quality evaluation information corresponding to the preprocessed acoustic signal, signal fusion is performed on each preprocessed acoustic signal in the preprocessed acoustic signal group to obtain the acoustic signal in the acoustic signal array that corresponds to the acquisition area.

7. The pipe loss identification method for water supply pipelines according to claim 6, characterized in that, The step of performing region correction on the target pipe loss region of interest based on the acoustic signal array and the target region feature map to obtain pipe loss description information includes: The acoustic signals in the acoustic signal array are encoded to obtain an acoustic signal feature map; Based on the fused image features, feature map sampling is performed on the acoustic signal feature map to obtain a sampled acoustic signal feature map, wherein the feature map size of the fused image features and the feature map size of the sampled acoustic signal feature map are the same; Local feature fusion is performed based on the acoustic signal feature map and the target region feature map to obtain multimodal fusion features; Based on the multimodal fusion features, fine-grained pipe loss localization is performed to obtain regional offset information; Based on the region offset information, the region of interest for the target pipe loss is updated to obtain pipe loss description information.

8. A pipe loss identification device for water supply pipelines, characterized in that, include: The determination unit is configured to determine the region of interest for initial pipe loss based on the water supply pipeline status data; The acquisition unit is configured to acquire infrared images and reflected wave signal sequences corresponding to the region of interest of the initial pipe loss, wherein the reflected wave signals in the reflected wave signal sequence are acquired by ground penetrating radar; The generation unit is configured to generate a humidity heatmap based on the reflected wave signal sequence; The pipe damage coarse localization unit is configured to perform coarse localization of pipe damage based on the infrared image and the humidity thermogram to obtain the target pipe damage region of interest and the target region feature map, wherein the target pipe damage region of interest is a sub-region of the initial pipe damage region of interest; The directional acquisition unit is configured to directionally acquire the acoustic signal array corresponding to the region of interest of the target pipe loss; The region correction unit is configured to perform region correction on the target pipe loss region of interest based on the acoustic signal array and the target region feature map to obtain pipe loss description information, wherein the pipe loss description information includes: pipe loss location and pipe loss type.

9. An electronic device, characterized in that, include: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 7.

10. A computer-readable medium, characterized in that, It stores a computer program thereon, wherein the computer program, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images