Pipeline inspection robot based on composite resolution acoustic perception and inspection method thereof

Abstract

The present application relates to the field of pipeline inspection, and particularly to a pipeline inspection robot based on composite resolution acoustic sensing and an inspection method thereof. The pipeline inspection robot comprises a power supply, a chassis, a probe and a controller. The probe is loaded on the chassis; the probe comprises a low-frequency sensing component and a high-frequency sensing component; the low-frequency sensing component performs wide-area scanning on the front of the chassis through a sweep signal of a preset frequency in an audio band; the high-frequency sensing component is used for ultrasonic imaging on a local area in front of the chassis through an ultrasonic signal of a preset frequency. The controller performs autonomous path planning and motion control of the robot according to the low-frequency sensing result, and lays a foundation for realizing more accurate high-frequency sensing. The high-frequency sensing result is used for correcting the path planning and task content of the robot, and also used for forming a more detailed inspection report. The collaborative acoustic sensing scheme provided by the present application can overcome the shortcomings of the existing scheme, and simultaneously realize high-precision, long-distance and cable-free automatic detection of internal defects of a pipeline.

Description

Technical Field

[0001] This invention relates to the field of pipeline inspection, and in particular to a pipeline inspection robot based on composite resolution acoustic perception, and the corresponding automatic pipeline inspection method. Background Technology

[0002] Urban drainage, water supply, gas, oil and gas, and industrial pipelines are mostly located underground or in enclosed environments. Problems such as misaligned pipe joints, bends, manholes, sediment, blockages, damage, and leaks directly impact public safety and maintenance costs. Existing detection methods mainly include video detection, electromagnetic detection, passive acoustic detection, and active ultrasonic detection.

[0003] Among these methods, camera-based inspection offers advantages such as intuitiveness and high resolution, but it relies on lighting, a clean field of view, and high-bandwidth data transmission. Long-term deployment results in significant power consumption, data volume, and increased workload for manual interpretation, often requiring cable crawlers and remote control. Electromagnetic inspection is suitable for detecting corrosion or cracks in metal pipes, but its applicability to non-metallic pipes such as plastics and concrete is poor. Passive acoustic inspection can be used for leak location, but its accuracy, pipe material adaptability, and defect type identification capabilities are limited. Traditional high-frequency ultrasonic inspection can generate high-resolution images, but it typically relies on liquid coupling, close-range scanning, or large arrays, making it difficult to directly undertake long-distance autonomous navigation tasks for pipeline robots.

[0004] Therefore, existing technologies generally have the following shortcomings: First, a single sensing method cannot simultaneously achieve long-distance navigation, close-range fine identification, and low-power autonomous operation; Second, although low-frequency long-distance acoustic signals are suitable for large-area scanning and candidate target localization, their spatial resolution is insufficient, making it difficult to directly determine the type of defect; Third, although high-frequency acoustic / ultrasonic arrays are suitable for fine structure imaging and detection, their effective range is relatively short, making it difficult to undertake long-distance navigation tasks independently. Summary of the Invention

[0005] To address the challenges of existing pipeline inspection solutions that struggle to achieve high precision, long-distance operation, and cable-free automated inspection, this invention provides a pipeline inspection robot based on composite resolution acoustic perception and its corresponding automated pipeline inspection method.

[0006] The technical solution provided by this invention is as follows: A pipeline inspection robot based on composite resolution acoustic perception includes a power supply, a chassis, a probe, and a controller. The power supply powers the remaining components, and the chassis moves along the target pipeline. The probe is mounted on the chassis and includes a low-frequency sensing component and a high-frequency sensing component. The low-frequency sensing component performs a wide-area scan in front of the chassis using a frequency sweep signal at a preset frequency within an audio band; the high-frequency sensing component performs ultrasonic imaging of a local area in front of the chassis using an ultrasonic signal at a preset frequency.

[0007] The controller is electrically connected to the chassis and the probe; and is used to perform the following tasks: (1) control the low-frequency sensing component to perform a wide-area scan in front of the chassis; identify the target type based on the effective signal and calculate the distance between it and the probe. (2) perform path planning and motion control on the chassis based on the scanning results of the low-frequency sensing component, so as to approach each target to be inspected in turn and make it at the optimal observation distance from the probe. (3) control the high-frequency sensing component to perform ultrasonic imaging on the target to be inspected; perform structural analysis and annotation on the target to be inspected based on the high-resolution imaging results, and output the inspection results; including: using the pipeline multi-view full-focusing algorithm to fuse multi-view information containing direct sound path, half-jump sound path and full-jump sound path to generate a three-dimensional ultrasonic image in the pipeline coordinate system; performing dynamic threshold binarization on the three-dimensional ultrasonic image to obtain a binary image; using any clustering method to cluster the non-zero pixels in the binary image along the pipeline axis and cross section to obtain several feature clusters; performing Radon transform on each feature cluster and extracting multiple feature parameters, including: axial distance d z axial length N z and angular non-zero proportion r nz ;according to d z , N z and r nz Classify target types and perform structural analysis.

[0008] As a further improvement of the present invention, the low-frequency sensing component includes a low-frequency sound source, a microphone array, and a signal acquisition board. The microphone array includes multiple sound signal receiving units arranged circumferentially along the cross-section of the pipe; the signal acquisition board is used to control the low-frequency sound source to emit a sweep frequency signal in the range of 0-20kHz to the pipe in front of the chassis, and simultaneously acquire the multi-channel echo signals received by the microphone array.

[0009] As a further improvement of the present invention, the high-frequency sensing component includes an air-coupled ultrasonic transducer array, an array controller, a transmit-receive switching circuit, and a data acquisition module. The ultrasonic transducer array includes multiple transducer elements; each transducer element is arranged circumferentially along the cross-section of the pipe in a hollow ring array; each transducer element constitutes several sets of transmitting elements and several receiving elements, with the transmitting and receiving elements arranged in pairs or at a predetermined angle, for transmitting ultrasonic signals to the inner wall of the pipe, interfaces, misalignments, blockages, bends, or tees, and receiving their reflected or scattered echoes; the array controller is electrically connected to the ultrasonic transducer array, the transmit-receive switching circuit, and the data acquisition module; the array controller generates a high-frequency excitation signal and controls the transmit-receive switching circuit according to a preset element selection sequence, so that the high-frequency excitation signal is applied to the selected transmitting element, while the echo signal received by the selected receiving element is transmitted to the data acquisition module.

[0010] As a further improvement of this invention, the pipeline inspection robot's inspection strategy for the target pipeline includes: first, using a low-frequency sensing component to perform a wide-area scan of the current section of the target pipeline within the visible range, identifying the pipeline structure and targets to be inspected contained therein. Then, based on the pipeline structure, the chassis performs path planning to generate an inspection path that can traverse all targets to be inspected within the current visible range, and generates an inspection list. Next, the robot is driven to approach each target to be inspected sequentially and performs ultrasonic imaging at the optimal observation distance. After completing the ultrasonic imaging task for all targets to be inspected in the inspection list, the aforementioned process is repeated to inspect the next section of the target pipeline until the inspection task for all pipelines within the target pipeline is completed.

[0011] As a further improvement of the present invention, when the scanning result of the low-frequency sensing component reflects that the front end of the target pipe includes an impassable pipe structure, the controller drives the vehicle to approach and perform ultrasonic imaging to verify it; if the verification is successful, the inspection task is stopped, and then the robot is driven to return along the original path and report the abnormal information.

[0012] As a further improvement of the present invention, when the imaging result of the high-frequency sensing component reflects that the distance between the target and the probe exceeds the optimal observation distance, the controller drives the vehicle to adjust its position and then re-performs ultrasonic imaging.

[0013] As a further improvement of the present invention, the pipeline structures identified in the scanning results of the low-frequency sensing component include empty pipes, blockages, aligned structures, inclined structures, branch pipe openings, and manhole openings; among them, blockages and manhole openings are impassable pipeline structures. Targets to be investigated include blockages, deposits, root intrusion, pipeline damage, inclined structures, and branch pipe openings.

[0014] As a further improvement of the present invention, the data processing process implemented by the controller based on the echo signal of the low-frequency sensing component includes: (1) Bandpass filtering, noise reduction and envelope extraction are performed on the original echo signal.

[0015] (2) Identify the echo peak position according to the preset threshold and extract the effective signal contained therein.

[0016] (3) Input each effective signal segment into the target classification model based on neural network pre-training to obtain the target classification result.

[0017] (4) Calculate the distance of the target to the probe corresponding to each effective signal segment based on the sound speed and echo propagation time.

[0018] As a further improvement of the present invention, the chassis is a motor-driven wheeled, tracked, support wheeled, or other type of robot suitable for movement within a tube.

[0019] As a further improvement of the present invention, the movement distance of the chassis is estimated based on the motor speed collected by the encoder, thereby realizing the spatial positioning of the robot in the pipeline.

[0020] As a further improvement of the present invention, the chassis also includes a gyroscope, which, based on the directional and acceleration information collected by the gyroscope, and in conjunction with the motor speed collected by the encoder, performs spatial positioning of the robot inside the pipeline.

[0021] As a further improvement of the present invention, the controller includes a data storage module, which is used to store the signals collected by the low-frequency sensing component and the high-frequency sensing component and the analysis results obtained after data processing.

[0022] As a further improvement of the present invention, the probe also includes a camera with a supplementary light, and the controller is used to control the camera to start so as to acquire on-site images of the specified target.

[0023] As a further improvement of the present invention, a rechargeable lithium battery is used as the power source.

[0024] This invention also includes an automated pipeline inspection method based on composite resolution acoustic perception, which employs the aforementioned pipeline inspection robot based on composite resolution acoustic perception and is used to perform segmented inspections of the target pipeline based on a known pipeline map. The automated pipeline inspection method provided by this invention includes the following steps: A spatial map of the target pipeline is obtained in advance, an inspection path is generated based on the spatial map, and multiple wide-area scanning nodes are set on the inspection path according to the visibility range of the low-frequency sensing component.

[0025] The robot starts from the starting point and arrives at the first wide-area scanning node. It uses low-frequency sensing components to perform a wide-area scan of the corresponding section of the target pipeline within the visible range, identifying the pipeline structure and the target to be investigated contained therein.

[0026] Based on the identified pipeline structure and the target to be inspected, the inspection path is analyzed for passability and optimized to generate an inspection path that can traverse all targets to be inspected within the current visible range, and an inspection list is generated.

[0027] The robot is controlled to approach each target in turn and perform ultrasonic imaging at the optimal observation distance. The imaging results are dynamically binarized and clustered to obtain several feature clusters. Three feature parameters of the feature clusters are calculated, including axial distance, axial length, and non-zero angular proportion. The targets are classified and their structures are analyzed based on the feature parameters.

[0028] After completing the ultrasonic imaging task of all targets to be inspected in the inspection list, the controller drives the robot to the next wide-area scanning node, repeats the above process to complete the inspection task of the next section of the target pipeline, until the end of the inspection path is reached.

[0029] The present invention has the following beneficial effects: This invention integrates a low-frequency acoustic signal-based sensing component and a high-frequency ultrasonic imaging component into a single probe, and designs a pipeline inspection robot. This solution combines the characteristics of the two acoustic sensing components to scientifically design the pipeline inspection strategy, achieving both large-scale autonomous navigation and local fine structure recognition. Specifically, low-frequency acoustics handles long-distance candidate feature localization and large-scale pipeline scanning, while high-frequency ultrasonic imaging handles local high-resolution 3D imaging and fine structure analysis. This composite-resolution acoustic sensing system overcomes the limitations of traditional single-resolution solutions in simultaneously achieving distance, resolution, and autonomous control.

[0030] Compared to traditional solutions, the present invention reduces the scale of data processing and equipment power consumption. For example, compared to continuous imaging or laser point cloud acquisition, the present invention performs high-frequency fine detection only in candidate areas and parameterizes the 3D image into a small number of classification features, making it suitable for low-power robot platforms. The inspection robot of the present invention can also automatically determine the moving distance, imaging position, continued inspection or return to base based on acoustic echo, ultrasonic images and classification results, reducing reliance on manual remote control and cables.

[0031] The air-coupled acoustic method employed in this invention can operate without relying on lighting or a clean visual environment, making it particularly suitable for drainage, sewage, and other pipes where visual inspection is difficult. This enhances the robot's autonomous inspection capabilities underground and allows it to adapt to low-light, dirty, or low-liquid-level environments. Furthermore, this invention supports the inspection of pipes made of any material. The number of array channels, acoustic frequency, imaging algorithm, classifier, and robot chassis in the two types of sensing components can all be flexibly adjusted according to the pipe diameter, material, defect type, and inspection task. Therefore, it offers optimized scalability and broader scenario adaptability. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the pipeline inspection robot based on composite resolution acoustic perception provided in Embodiment 1 of the present invention.

[0033] Figure 2 This is a schematic diagram illustrating the working principle of the pipeline inspection robot inside the pipeline in Embodiment 1 of the present invention.

[0034] Figure 3 This is a schematic diagram of the probe part of the pipeline inspection robot in Embodiment 1 of the present invention.

[0035] Figure 4 The logic block diagram for the pipeline inspection robot in embodiment 1 to achieve autonomous perception and motion control is shown below.

[0036] Figure 5 This is a schematic diagram of the working scenario of the pipeline inspection robot provided by the present invention during a test experiment.

[0037] Figure 6 This is the pipeline under test in a typical test scenario for the test experiment.

[0038] Figure 7 To test the echo signals collected by the low-frequency sensing component for different pipe structures in the experiment.

[0039] Figure 8 To test the ultrasound imaging results of the high-frequency sensing component for different targets in the experiment.

[0040] Figure 9 These are characteristic parameter images of different types of targets analyzed from ultrasound images during the test experiment.

[0041] The diagram is marked as follows: 1. Chassis; 2. Power supply; 4. Camera; 30. Base plate; 31. Low-frequency sound source; 32. Microphone array; 33. Ultrasonic transducer; 100. Controller. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the specification of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "or / and" as used herein includes any and all combinations of one or more of the associated listed items.

[0044] Example 1 To address the challenge of existing pipeline inspection equipment in harsh, narrow, low-light, or low-liquid-level pipeline environments, which struggle to simultaneously achieve long-distance spatial positioning, close-range 3D imaging, defect classification, and automatic detection, this embodiment provides a pipeline inspection robot based on composite resolution acoustic perception. This solution integrates acoustic perception components with different resolutions onto a single device. It then utilizes low-frequency acoustic signals for large-scale scanning and coarse positioning of the pipeline; combines the scanning and positioning results for path planning and fine-tuning of the observation distance; and finally, employs high-frequency ultrasonic sensing technology for close-range, fine-structure perception of localized areas identified in the initial scan. This solution leverages the characteristics of different resolution acoustic perception components in terms of propagation distance and data processing scale to efficiently achieve high-precision automatic detection of targets within pipelines, achieving a balance between detection accuracy and equipment energy consumption. This lays the foundation for autonomous navigation, precise positioning, active obstacle avoidance, and more refined defect identification within pipelines.

[0045] Specifically, such as Figure 1 As shown, the pipeline inspection robot based on composite resolution acoustic perception provided in this embodiment includes a power supply 2, a chassis 1, a probe, and a controller 100. Figure 2 As shown, in this embodiment, the pipeline inspection component mainly includes two functions: First, it moves within the pipeline via chassis 1, autonomously planning its path and avoiding obstacles during movement to traverse the targets within the pipeline to be inspected. Second, during the movement of chassis 1, it utilizes probes and controller 100 mounted on it to perform multi-dimensional acoustic perception and structural analysis of targets at various locations within the pipeline to identify pipeline defects and generate corresponding inspection reports. The defect analysis results recorded in the inspection reports will lay the foundation for subsequent pipeline maintenance.

[0046] Based on the aforementioned functional requirements, the chassis 1 in this embodiment is used to carry various functional components and move along the target pipeline. In practical applications, the chassis 1 in this embodiment can be a motor-driven wheeled, tracked, support-wheeled, or any other type of robot suitable for movement within a pipe. The power supply 2 is used to power the remaining components. In practical applications, the power supply 2 can be a high-capacity rechargeable lithium battery, or other battery packs with higher safety and higher power density. In this embodiment, both the chassis 1 and the electronic equipment loaded on it are powered by the configured battery, which reduces the dependence on the external power supply 2, thereby achieving cable-free automatic inspection within the pipeline. Furthermore, considering that the chassis 1 can usually only move along the axial direction of the pipeline inside the pipeline, this embodiment can also use the motor speed collected by the encoder to estimate the movement distance of the chassis 1, thereby achieving spatial positioning of the inspection robot within the pipeline. Of course, in practical applications, in order to achieve more accurate spatial positioning of the inspection robot inside the pipeline, this embodiment can also integrate sensors such as gyroscopes on the chassis 1; thereby enabling spatial positioning of the chassis 1 inside the pipeline based on the direction and acceleration information collected by the gyroscope, in conjunction with the motor speed collected by the encoder.

[0047] In this embodiment, the probe is mounted on chassis 1; in practical applications, such as... Figure 3 As shown, the probe includes a base plate 30 whose shape and size match the cross-section of the pipe under test, and a low-frequency sensing component and a high-frequency sensing component mounted on the base plate 30. For example, when the pipe under test is a circular pipe, in order to achieve a larger sensing field of view while ensuring passability, the base plate 30 can be a disc with a size slightly smaller than the inner diameter of the pipe under test. In practical applications, the base plate 30 can be vertically mounted on the chassis 1 at the front end corresponding to the direction of movement. In the probe provided in this embodiment, the low-frequency sensing component is used to perform a wide-area scan in front of the chassis 1 using a frequency sweep signal of a preset frequency in the audio band. The high-frequency sensing component is used to perform ultrasonic imaging of a local area in front of the chassis 1 using an ultrasonic signal of a preset frequency. Of the two acoustic sensing components, the former has a lower resolution but a longer detection distance and lower power consumption; based on this characteristic, this embodiment uses it to quickly detect and coarsely locate various structural targets inside the pipe before approaching them. The latter has a higher resolution but a smaller field of view, higher power consumption, and a relatively larger data processing scale; therefore, this embodiment uses it to perform close-range fine sensing and verification of targets detected by the former.

[0048] In this embodiment, the low-frequency sensing component includes a low-frequency sound source 31, a microphone array 32, and a signal acquisition board. For example... Figure 3As shown, the microphone array 32 includes multiple acoustic signal receiving units arranged circumferentially along the cross-section of the pipe. The signal acquisition board is used to control the low-frequency sound source 31 to emit a sweep frequency signal in the range of 0-20kHz to the pipe in front of the chassis 1, and simultaneously acquire the multi-channel echo signals received by the microphone array 32. For example, in a typical scheme, the low-frequency sound source 31 can be a loudspeaker, which is arranged in the center of the probe substrate 30, and the microphone array 32 includes six microphones evenly arranged in a ring around the loudspeaker. In practical applications, the low-frequency sensing component of this embodiment can combine the specific operating parameters such as the size and material of the pipe, as well as the cutoff frequency of the acoustic mode, to reasonably configure the signal frequency band of the sweep frequency signal emitted by the sound source to achieve the best detection effect. For example, in a typical municipal drainage pipe, the sound source can emit a swept frequency signal in the range of 300 Hz to 600 Hz. This frequency band is just below the cutoff frequency of the higher-order modes in a 300 mm pipe, preferentially exciting the (0,0) mode of an approximate plane wave, thereby achieving long-distance propagation in the pipe axis and generating echo signals when encountering locations where there are significant changes in acoustic impedance, such as joints, bends, manholes, blockages, or pipe ends.

[0049] Accordingly, the high-frequency sensing component in this embodiment includes an air-coupled ultrasonic transducer array, an array controller, a transmit-receive switching circuit, and a data acquisition module. The ultrasonic transducer array includes multiple transducer elements; such as... Figure 3 As shown, each transducer element is arranged in a hollow ring array along the circumferential direction of the pipe cross-section; a clearance space is formed in the middle of the array for low-frequency acoustic units or other structures to pass through. For example, each transducer element can be arranged on the substrate 30 corresponding to the outer periphery of the microphone array 32. Each transducer element constitutes several sets of transmitting elements and several receiving elements, with transmitting and receiving elements arranged in pairs or at predetermined angles. The transmitting elements are used to transmit ultrasonic signals to the inner wall of the pipe, interfaces, misalignments, blockages, bends, or tees, and receive their reflected or scattered echoes through the receiving elements. The array controller is electrically connected to the ultrasonic transducer array, the transmitting-receiving switching circuit, and the data acquisition module; the array controller generates a high-frequency excitation signal and controls the transmitting-receiving switching circuit according to a preset element selection sequence, so that the high-frequency excitation signal is applied to the selected transmitting element, while the echo signal received by the selected receiving element is transmitted to the data acquisition module.

[0050] In this embodiment, the transmit-receive switching circuit may include an analog multiplexer or a multi-channel transmit-receive switch for channel selection among multiple transmit array elements and multiple receive array elements. The data acquisition module includes a receive amplification circuit, a filter circuit, and an analog-to-digital conversion circuit for amplifying, filtering, and digitizing the echo signals output by the receive array elements. The array controller may employ a single-channel ultrasonic controller and a multiplexer board to achieve full matrix acquisition of multiple array elements, or it may employ a multi-channel ultrasonic chip for parallel acquisition. Based on this, the array controller obtains full matrix acquisition data corresponding to multiple transmit-receive channel combinations by sequentially selecting different combinations of transmit and receive array elements. Alternatively, the array controller may employ a multi-channel ultrasonic chip to acquire multiple transmit-receive channels in parallel. The full matrix acquisition data is used to generate local three-dimensional ultrasonic images of the pipeline based on a multi-view full-focus imaging algorithm.

[0051] In the typical scheme provided in this embodiment, the high-frequency sensing component can employ 24 air-coupled ultrasonic transducers 33 with a center frequency of approximately 40 kHz. The 24 ultrasonic transducers 33 are combined into 12 pairs of transmit-receive array elements and arranged in a clock-face hollow circular array. Unlike the low-frequency sensing unit, which can achieve wide-area scanning over distances exceeding 100 m, the high-frequency sensing component in this embodiment has a relatively narrow field of view. It can be used for detailed three-dimensional imaging of structures such as pipe inner walls, interfaces, misalignments, blockages, bends, and tees within a preferred observation distance of 40 cm to 100 cm.

[0052] Based on the chassis 1 and its load-bearing various detection devices, the controller 100 in this embodiment serves as the control center for the coordinated control of various components with different functions within the pipeline inspection robot. This achieves a perfect closed loop within the single robot, encompassing "large-area acoustic scanning + fine-structure perception of candidate areas + parametric classification + motion control." This allows the pipeline inspection robot to autonomously perform pipeline inspection tasks underground without relying on manual control.

[0053] Specifically, in this embodiment, the controller 100 is electrically connected to the chassis 1 and the probe; and the following tasks are performed based on the control program built into the controller 100: (1) Control the low-frequency sensing component to perform a wide-area scan in front of the chassis 1; preprocess the echo signal received by the low-frequency sensing component to extract the effective signal contained therein; identify the target type corresponding to each effective signal through a target classification model based on a neural network pre-training, and calculate the distance between it and the probe according to the acoustic propagation characteristics. (2) Perform path planning and motion control on the chassis 1 according to the scanning results of the low-frequency sensing component; so as to approach each target to be investigated in turn, and make the probe and each target to be investigated at the optimal observation distance. (3) When the probe and any target to be investigated are at the optimal observation distance; control the high-frequency sensing component to perform ultrasonic imaging on the target to be investigated; perform structural analysis and annotation on the target to be investigated according to the high-resolution imaging results, and output the inspection results.

[0054] Therefore, even if the pipeline inspection robot in this embodiment is unfamiliar with the spatial distribution of underground pipelines, it can first obtain a "spatial map" of a local area through wide-area scanning of the low-frequency sensing component. Then, based on the results of the wide-area scanning, it can plan the robot's inspection trajectory within the pipeline and formulate a high-frequency sensing imaging scheme for each target to be inspected. This achieves autonomous "map-less inspection" of pipelines with complex spatial topologies. Specifically, as... Figure 4 As shown, the pipeline inspection robot's "map-less inspection" strategy for the target pipeline includes: first, using a low-frequency sensing component to perform a wide-area scan of the current segment of the target pipeline within the visible range, identifying the pipeline structure and the targets to be inspected. Then, based on the pipeline structure, the chassis 1 performs path planning to generate an inspection path that can traverse all targets to be inspected within the current visible range, and generates an inspection list. Next, the robot is driven to approach each target to be inspected sequentially and performs ultrasonic imaging at the optimal observation distance. After completing the ultrasonic imaging task for all targets to be inspected in the inspection list, the aforementioned process is repeated to inspect the next segment of the target pipeline until the inspection task for all pipelines within the target pipeline is completed.

[0055] During the execution of the aforementioned inspection tasks, the planned inspection path is not static but dynamically adjusted based on the specific detection results. For example, when the low-frequency sensing component's scan results indicate that the target pipe's front end includes an impassable pipe structure, the controller 100 will drive the robot to approach and perform ultrasonic imaging to verify this. If the verification is successful, the inspection task is terminated, and the robot returns along the original path, reporting the anomaly. If the verification fails, the node is modified to a passable structure, and an updated inspection path and inspection list are generated.

[0056] In this embodiment, low-frequency acoustic signals can propagate for hundreds of meters within the pipeline structure, detecting structures such as pipeline interfaces, tees, manholes, diameter changes, and blockages. Furthermore, low-frequency sound waves can bypass bends and tees to continue propagating. Therefore, the received low-frequency sound wave signal contains not only the structure and target objects of the main pipeline but also structural information of branch pipelines. Currently, this embodiment primarily utilizes low-frequency sensing components for a rough scan of the pipeline structure over long distances, providing the pipeline robot with approximate target location information and a certain degree of category information for the pipeline structures ahead. Objects within the pipeline identified by low-frequency acoustic sensing in this embodiment can be divided into two categories: passable structures (interfaces, tees, diameter changes) and impassable structures (blockages, manholes). For structural integrity checks, such as the integrity of pipeline interfaces (intact connection or deflection), the integrity of tees, and the size of blockages, high-frequency sensing components perform detailed detection.

[0057] In practical applications, when multiple targets with different structures exist in a target pipeline, the effective signals in the acquired low-frequency echoes will vary significantly. For example, when structures such as bends, deflections, or manholes appear ahead of the pipeline, or when there are blockages inside the pipeline, the peak value and envelope of the echo signal will change significantly. Therefore, the above-mentioned characteristics in the echo signals acquired by the low-frequency sensing component can be combined to quickly identify different targets. In this embodiment, the pipeline structures that can be identified in the scanning results of the low-frequency sensing component include empty pipes, blockages, aligned structures, inclined structures, branch pipe openings, and manhole openings. Among them, blockages and manhole openings are impassable pipeline structures. In the detection results of the low-frequency sensing component, the targets that need to be further verified by the high-frequency sensing component mainly include blockages, deposits, root intrusion, pipeline damage, inclined structures, and branch pipe openings.

[0058] Specifically, in this embodiment, the low-frequency sensing component sends the acquired echo signal to the controller 100 during each round of wide-area scanning, and the controller 100 performs data processing and analysis locally. Specifically, the process by which the controller 100 processes the data acquired by the low-frequency sensing component includes the following steps: (1) Bandpass filtering, noise reduction and envelope extraction are performed on the original echo signal.

[0059] It should be noted that, in order to improve the sensitivity and detection accuracy of the low-frequency sensing component, this embodiment can also pre-scan the pipeline in its defect-free state when it first starts to be put into service, and use the echo signal obtained from the pre-scan as the background signal for the subsequent actual detection process. Then, the background signal is used to denoise the detection signal obtained in the actual detection process, thereby highlighting the signal components of abnormal structures contained therein.

[0060] (2) Identify the echo peak position according to the preset threshold and extract the effective signal contained therein.

[0061] In this embodiment, the echo signal collected by the low-frequency sensing component is a time-domain signal. The presence of any target or pipeline structure in this signal will cause a local morphological change in that segment of the echo signal. Therefore, this segment of the echo signal can be extracted as a valid signal. In practical applications, this embodiment can select the extracted echo peak position based on a set threshold. The threshold can be set as a predetermined proportion of the echo amplitude at the pipeline end or inspection well, such as one-tenth, or it can be adaptively adjusted according to the ambient noise.

[0062] (3) Input each effective signal segment into the target classification model based on neural network pre-training to obtain the target classification result.

[0063] The differences in echo signals between different pipe structures have been described previously. Therefore, in order for the pipe inspection robot to accurately identify different pipe structures based on signal characteristics and lay the foundation for subsequent path planning and high-resolution imaging, this embodiment incorporates a pre-trained lightweight target classification model in the controller 100. Simply inputting a valid signal into the target classification model allows it to quickly output the target type corresponding to the current signal.

[0064] (4) Calculate the distance of the target to the probe corresponding to each effective signal segment based on the sound speed and echo propagation time.

[0065] In the original echo signal, due to the influence of the signal propagation process, the effective signal corresponding to the target at different distances will appear in different time periods of the echo signal. Therefore, in this embodiment, the echo propagation time of the effective signal is multiplied by the propagation speed of the audio in the current detection scene and divided by 2 to obtain the spatial distance between the target and the probe corresponding to the effective signal.

[0066] After completing low-frequency sensing of the target section of the pipeline and planning the inspection trajectory and list, the high-frequency sensing phase begins. In this phase, the controller 100 can select the optimal ultrasonic frequency to detect different targets based on the actual working conditions. During high-frequency sensing, different pipeline structures, such as interfaces, tees, and reducers, appear only as annular gaps in the high-frequency ultrasonic image. If the ultrasonic echo signal reflection energy distribution of the gap structure is uniform in the 3D ultrasonic image, the pipeline structure can be considered intact, meaning the interface or structure is neatly aligned with the pipeline (aligned structure). If the ultrasonic echo signal reflection energy distribution of a gap structure is uneven along the pipeline axis in the 3D ultrasonic image, the interface structure is considered problematic, indicating interface deflection, which can be classified as an misaligned structure. Another aspect that needs to be identified from the high-frequency ultrasonic image information is the structure of blockages, which is used to assess the pipeline's passability and determine whether the robot needs to stop its inspection and return. The quantitative analysis results of pipeline damage or blockages based on the ultrasonic image analysis during the high-frequency sensing phase can also be used by subsequent maintenance personnel to assess the severity of the problem.

[0067] In this embodiment, the data processing procedure performed by the controller 100 based on the image data acquired by the high-frequency sensing component includes the following: (i) The multi-view full-focusing algorithm of the pipeline is adopted to fuse multi-view information including direct acoustic path, semi-jump acoustic path and full-jump acoustic path to generate a three-dimensional ultrasound image in the pipeline coordinate system.

[0068] Once the pipeline inspection robot moves to the corresponding position based on the scanning results of the low-frequency sensing component, it performs full-matrix acquisition via the ultrasonic transducer array. For the acquired imaging data, the controller 100 can employ a multi-view full-focusing algorithm to fuse information from multiple perspectives, including direct acoustic paths, semi-skipped acoustic paths, and full-skipped acoustic paths, to generate a 3D image in the pipeline coordinate system. Different perspective images can be fused using Tippet fusion, Fisher fusion, or other statistical fusion methods to reduce the impact of the limited number of array elements and single-view artifacts on imaging quality.

[0069] The multi-view full-focus imaging method for pipelines used in this embodiment includes the following steps: Obtain the geometric model of the object under test; and establish an acoustic propagation model for detecting the object under test using an ultrasonic phased array based on the state parameters of the object under test.

[0070] The transceiver array pairs in the ultrasonic phased array are traversed, and each sound ray emitted and propagated at a preset angular interval is tracked; each candidate path that meets the preset constraints is recorded; and the path characteristics of each candidate path are generated, including: number of reflections, path length, propagation time, incident angle, reflection angle, boundary intersection, reflection coefficient, and path energy.

[0071] Perform transmit / receive array element matching on each candidate path; and filter the matched candidate paths using the following formula to obtain the set of effective paths. : ; In the above formula, p The voxel to be imaged is represented by i; i and j represent the indices of the transmitting and receiving array elements. Indicates the transmit and receive array elements Candidate paths k Path energy; Indicates the transmit and receive array elements In voxels p Energy along a straight path at a given point; η The preset energy ratio threshold; N k Indicate candidate path k The number of reflections; N max This indicates the preset maximum number of reflections; Indicates belonging to the transmit / receive array element pair Candidate paths k The spread time.

[0072] Three-dimensional ultrasound imaging of the object under test is performed by combining the effective path set, including: first, generating the path weight and focusing delay of the effective path; then, delaying and sampling the A-scan signals acquired by each transceiver array element according to the effective path corresponding to each voxel and superimposing them to obtain the image intensity of a specified number of reflections; finally, fusing the image intensities of different number of reflections or different path families to obtain the three-dimensional ultrasound image of the object under test.

[0073] The path weight of any valid path in this embodiment Based on the path energy of each effective path The generation and calculation formula is: ; In the above formula, p Indicates the voxel to be imaged; P ( p ) indicates the participating voxels p The effective path categories or viewpoint set for image fusion; ε is a positive constant to prevent the denominator from being zero, which is a very small positive number. The obtained path weights can be used for subsequent multipath image fusion.

[0074] Focusing delay of any valid path Based on the propagation time of the current path The generation and calculation formula is: ; In the above formula, This refers to the system trigger delay, zero-point acquisition delay, or calibration delay. In actual imaging, for the same voxel p and the same transceiver array pair... There can be k different types of optional paths with corresponding effective delays.

[0075] In the three-dimensional ultrasound imaging stage of this embodiment, according to the first k Voxels generated by the effective path of secondary reflection p The image intensity satisfies the following formula: ; In the above formula, and These represent the number of transmitting and receiving elements of the ultrasonic phased array, respectively. This represents the sampled value of the A-scan signal at the predicted propagation time.

[0076] If imaging is performed separately for different reflection counts or different path families, the final three-dimensional image will be... It can be generated by the following formula: ; in, F The preset image fusion method is indicated. In this embodiment, the fusion method of the three-dimensional ultrasound images of the object under test adopts weighted summation, maximum value, Tippet statistical fusion, Fisher statistical fusion or other confidence fusion methods.

[0077] The expression for weighted summation and fusion is as follows: ; In the above formula, This represents the image intensity at voxel p generated based on weighted fusion; P ( p ) indicates the participating voxels p Effective path categories or viewpoint sets for image fusion; Indicates the first k Class path image in voxels p The weighted summation and fusion weights are calculated based on the weighted summation and fusion weights at the specified points. k The path energy, signal-to-noise ratio, background noise variance, or path confidence of the path are determined.

[0078] The expression for Tippet fusion is: ; In the above formula, Indicates voxels generated based on weighted fusion p Image intensity at that location; Indicates the first k Class path image in voxels p Tippet fusion weights at the location.

[0079] The expression for Fisher fusion is: ; In the above formula, This indicates voxels generated based on Fisher fusion. p Image intensity at that location; Indicates the first k Class path image in voxels p Fisher fusion weights at the location.

[0080] (ii) The three-dimensional ultrasound image is processed by dynamic thresholding to obtain a binary image.

[0081] To accommodate the computational constraints of the pipeline inspection robot when processing data locally, this embodiment uses dynamic thresholding binarization to obtain binary images from the obtained 3D ultrasonic images. In practical applications, the dynamic threshold used for binarization can be determined based on pre-measured empty pipe noise distribution, image normalization intensity, and distance-dependent signal-to-noise ratio. Furthermore, similar to the low-frequency sensing component mentioned earlier, the high-frequency sensing component can also pre-image various locations within the non-destructive pipeline before performing actual inspection on any given pipeline, using the pre-image results as background signals to correct the ultrasonic image data detected in each round.

[0082] (iii) Use any clustering method to cluster the non-zero pixels in the binary image along the pipe axis and cross-section to obtain several feature clusters.

[0083] (iv) Perform Radon Transform (RT) on each feature cluster and extract multiple feature parameters, including: axial distance. d z axial length N z and angular non-zero proportion r nz ; (v) According to d z , N z and r nz Classify target types and perform structural analysis.

[0084] In this embodiment, to enable the pipeline inspection robot to directly identify the specific type and detailed structure of the target objects contained in the ultrasonic images locally, this embodiment abandons the traditional machine vision-based algorithm scheme. Instead, it first reduces the dimensionality of the ultrasonic images to binary images, and then analyzes the statistical features of the targets to be investigated in the binary images. Specifically, this embodiment first clusters the non-zero pixels in the binary images along the pipeline axis and cross-section, preferably using clustering methods such as DBSCAN to separate adjacent features. Then, it extracts three key feature parameters of the clustered feature clusters; finally, it classifies the targets based on the feature parameters.

[0085] Specifically, among the three characteristic parameters in this embodiment, the axial distance... d z This refers to the distance between the center of the feature cluster and the probe; while the axial length... N z This refers to the maximum length of the feature cluster along the pipe axis. Angular non-zero ratio. r nz This refers to the angular proportion of the distribution area of ​​the feature cluster in the radial direction of the pipe. For example, if sediment accumulates in the lower section of the pipe and occupies half of the circumferential space of the pipe, then the non-zero angular proportion is 180° / 360°=0.5.

[0086] The three types of indicators exhibit different patterns for different types of targets. Therefore, the type of target can be distinguished by the range of values ​​for relevant parameters. For example, blockage-type features typically form concentrated high-intensity pixels in a localized area of ​​the cross-section. r nz The value is relatively low; pipe structure features such as joints, tees, elbows, or pipe ends are usually distributed circumferentially along the pipe boundary. r nz The value is high; misaligned or tilted structures will have a higher axial length. N z It exhibits characteristics different from the aligned structure.

[0087] To achieve high imaging quality, the high-frequency sensing component in this embodiment always moves to a position at the optimal observation distance for each target before activating the ultrasonic transducer matrix 33 to acquire data. Therefore, the target is always positioned optimally within the corresponding field of view of the ultrasonic image. After the controller 100 acquires the imaging results from the high-frequency sensing component, it further analyzes the axial distance... d zThe system checks whether the distance between the target and the probe exceeds the optimal observation distance, for example, less than 40 cm or greater than 100 cm. If so, the controller 100 calculates the position correction and drives the chassis forward or backward to adjust the position, then performs ultrasound imaging and analysis again. This ensures that the high-frequency sensing unit always stays within the optimal observation distance when detecting each target.

[0088] In addition to classifying the target based on the three types of features mentioned above, decisions can also be made based on other feature information. For example, if the effective pixel count of a binary ultrasound image... N p Below a predetermined threshold, for example N p If the number of valid pixels is ≤125, the current location is determined to have no identifiable features or be an empty pipe, and the device moves to the next candidate location according to the inspection list. If the number of valid pixels exceeds the threshold, RT parameterization and classification are performed.

[0089] During pipeline inspection, the robot makes different decisions based on different inspection results to complete the inspection task more efficiently and safely. For example, if the classification result is a blockage or a suspected impassable obstacle, the robot will stop moving forward, record the distance and type of the obstacle, and return or wait for manual intervention. If the classification result is a structure and the structure corresponds to a pipe end or inspection well, the robot records the destination and returns. If the classification result is a structure but not the destination, the robot records the structure and continues to probe other targets on the inspection list along the preset path. If low-frequency acoustics detects an echo at the end of a branch pipe, while ultrasonic imaging shows that the main pipe is currently empty, it is determined that the echo may have originated from the branch pipe or a lateral propagation path, avoiding misjudging it as a main pipe obstacle.

[0090] In the practical application of this embodiment, the controller 100 includes a data storage module. This module stores the signals collected by the low-frequency and high-frequency sensing components and the analysis results obtained after data processing. This data will serve as the data source for generating subsequent inspection reports. In a further optimized version of this embodiment, to perform more detailed inspections of certain key sections or targets, the probe also includes a camera 4 with integrated supplementary lighting. The controller 100 controls the camera 4 to start according to preset trigger conditions and then collect on-site images of each target to be tested. For example, when the controller 100 detects a break in a pipeline based on the high-frequency sensing results, it will activate the camera 4 to collect supplementary on-site images.

[0091] In summary, the pipeline inspection robot based on composite resolution acoustic perception provided in this embodiment utilizes the differences in propagation distance, wavelength, signal-to-noise ratio, data volume, and structural representation capability of acoustic signals at different resolution levels within the pipeline to form composite resolution perception. Low-frequency acoustic waves have longer wavelengths and lower attenuation, allowing them to propagate several meters or more along the pipeline and generate detectable echoes from locations with impedance changes such as bends, interfaces, manholes, and blockages. This embodiment uses these waves for pre-scanning of a large area, candidate feature discovery, and robot navigation. High-frequency air-coupled acoustic / ultrasound waves have shorter wavelengths and higher spatial resolution, but a shorter effective detection range. Therefore, the inspection robot in this embodiment approaches candidate targets guided by the low-frequency large-area scan results and places them within the optimal observation distance; then, it performs high-frequency fine structure perception. The high-frequency image obtains the spatial distribution of the target through multi-view full-focusing and image fusion, and then compresses the three-dimensional image into a small number of feature parameters through dynamic binarization, clustering, and RT parameterization.

[0092] The robot's motion control within the pipeline uses low-frequency acoustic distance as the initial movement basis and high-frequency image-extracted axial distance as the position correction basis, enabling the robot to acquire stable images within a suitable imaging window. The classification results of the pipeline structure and the target to be inspected obtained in these two stages are used to determine whether to control the robot to move forward, adjust its position, or return. They are also used to generate the final inspection report. This embodiment thus forms a closed-loop autonomous inspection system driven by composite resolution acoustic perception.

[0093] Example 2 In Example 1 above, a pipeline inspection robot based on composite resolution acoustic perception was provided, and a strategy for autonomous "map-less inspection" of unknown pipelines based on this robot was introduced. Building upon this, this example further provides a new automatic pipeline inspection method based on composite resolution acoustic perception. This method employs the pipeline inspection robot based on composite resolution acoustic perception as described in Example 1, and uses it to perform segmented inspections of target pipelines based on a known pipeline map. That is, it achieves "map-based inspection." Specifically, the automatic pipeline inspection method provided in this example includes the following steps: S1: Pre-acquire the spatial map of the target pipeline, generate an inspection path based on the spatial map, and set multiple wide-area scanning nodes on the inspection path according to the visibility range of the low-frequency sensing component.

[0094] In this embodiment, the wide-area scanning nodes need to be set in combination with the spatial distribution of the pipeline and the visibility range of the low-frequency sensing components, so as to ensure that the visibility range of all wide-area scanning nodes can cover all sections of the pipeline under test and avoid forming blind spots.

[0095] S2: Drive the robot from the starting point to the first wide-area scanning node, and use the low-frequency sensing component to perform a wide-area scan of the corresponding section of the target pipe within the visible range, and identify the pipe structure and the target to be investigated contained therein.

[0096] S3: Based on the identified pipeline structure and the target to be inspected, perform passability analysis and path optimization on the inspection path to generate an inspection path that can traverse all targets to be inspected within the current visible range, and generate an inspection list.

[0097] In this embodiment, the spatial map of the pipeline is mainly used to guide the direction and sequence of the overall inspection task. For each section, the robot's inspection path in that section is still determined by the low-frequency perception results and dynamically adjusted after verification based on the high-frequency perception results.

[0098] S6: Control the robot to approach each target in turn and perform ultrasonic imaging at the optimal observation distance. Dynamically binarize and cluster the imaging results to obtain several feature clusters; calculate three feature parameters for each feature cluster, including axial distance, axial length, and non-zero angular proportion; classify and analyze the targets based on the feature parameters.

[0099] S7: After completing the ultrasonic imaging task of all targets to be inspected in the inspection list, the controller 100 drives the robot to the next wide-area scanning node, repeats the above process to complete the inspection task of the next section of the target pipeline, until the end of the inspection path is reached.

[0100] Test Experiment To verify the feasibility and performance advantages of the technical solution provided by this invention, technicians manufactured, as follows: Figure 5 The sample of the pipeline inspection robot shown is illustrated. And in a... Figure 6 The test was conducted on a PVC pipe with a length of 18m and a diameter of 300mm, as shown. This pipe includes typical straight pipe structures, T-turn structures, deflection structures, and access well structures.

[0101] In the probe of this pipeline inspection robot, the low-frequency sound source is a loudspeaker, which can emit a sweeping signal in the range of 300 Hz to 600 Hz. The microphone array is a six-microphone array arranged circumferentially along the cross-section of the pipeline. The center frequency of the air-coupled ultrasonic transducer is approximately 40 kHz; the probe uses 12 sets of transmit-receive array elements consisting of 24 ultrasonic transducers.

[0102] In the above test scenario, the effective signals of the echo signals collected by the low-frequency sensing component in the pipeline inspection robot at various locations are as follows: Figure 7As shown, in addition to the objects being inspected, the pipeline also contains targets composed of different types of blockages. Furthermore, after planning the robot's inspection path based on the detection results of the low-frequency sensing component, the high-frequency sensing component obtains ultrasonic images of each target inside the pipeline, as shown... Figure 8 As shown. This experiment performed binarization and cluster analysis on the ultrasonic structures of some targets to be investigated, and the analysis results of the three types of feature parameters are as follows. Figure 9 As shown. Figure 9 The data displayed is 300. Based on the data in the figure, different shapes of blockages, aligned / unaligned pipe interfaces, aligned / unaligned pipe tees, aligned / unaligned pipes, etc. can be identified.

[0103] Finally, in the test experiment, the present invention achieved an average standard deviation of about 8 cm and a maximum positioning error of about 25 cm in 30 autonomous inspections within an 18 m pipeline network. It can also classify features such as interfaces, misaligned structures, empty pipes and blockages. Therefore, it can meet the performance requirements of products in actual inspection scenarios and achieve similar accuracy to traditional visual inspection solutions under cableless and automatic detection conditions.

[0104] The above-described embodiments are merely one implementation of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A pipeline inspection robot based on composite resolution acoustic perception, characterized in that, It includes: power supply, Chassis, which is used to travel along the target pipeline; The probe is mounted on a chassis; the probe includes a low-frequency sensing component and a high-frequency sensing component based on sound waves; the low-frequency sensing component performs a wide-area scan of the area in front of the chassis using a frequency sweep signal of a preset frequency in the audio band; the high-frequency sensing component is used to perform ultrasonic imaging of a local area in front of the chassis using an ultrasonic signal of a preset frequency. The controller is electrically connected to the chassis and the probe; and is used to: (I) control the low-frequency sensing component to perform a wide-area scan in front of the chassis; identify the target type based on the effective signal and calculate the distance between it and the probe; (II) perform path planning and motion control of the chassis based on the scanning results of the low-frequency sensing component, so as to approach each target to be investigated in turn and make it at the optimal observation distance from the probe. (III) Control the high-frequency sensing component to perform ultrasonic imaging on the target under investigation, and perform structural analysis and annotation on the target under investigation based on the imaging results, including: using a pipeline multi-view full-focusing algorithm to fuse multi-view information including direct acoustic path, semi-jump acoustic path, and full-jump acoustic path to generate a three-dimensional ultrasonic image in the pipeline coordinate system; performing dynamic threshold binarization on the three-dimensional ultrasonic image to obtain a binary image; using any clustering method to cluster the non-zero pixels in the binary image along the pipeline axis and cross-section to obtain several feature clusters; performing Radon transform on each feature cluster and extracting multiple feature parameters, including: axial distance. d z axial length N z and angular non-zero proportion r nz ;according to d z , N z and r nz Classify target types and perform structural analysis.

2. The pipeline inspection robot based on composite resolution acoustic perception as described in claim 1, characterized in that: The low-frequency sensing component includes a low-frequency sound source, a microphone array, and a signal acquisition board; The microphone array includes multiple acoustic signal receiving units arranged circumferentially along the cross-section of the pipe; the signal acquisition board is used to control the low-frequency sound source to send a sweep frequency signal in the range of 0-20kHz to the pipe in front of the chassis, and simultaneously acquire the multi-channel echo signals received by the microphone array.

3. The pipeline inspection robot based on composite resolution acoustic perception as described in claim 2, characterized in that: The high-frequency sensing component includes an air-coupled ultrasonic transducer array, an array controller, a transmit-receive switching circuit, and a data acquisition module. The ultrasonic transducer array includes multiple transducer elements; each transducer element is arranged in a hollow ring array along the circumferential direction of the pipe cross-section; each transducer element constitutes several sets of transmitting elements and several receiving elements, and the transmitting elements and receiving elements are arranged in pairs or at a predetermined angle relationship, for transmitting ultrasonic signals to the inner wall of the pipe, interface, misalignment, blockage, elbow or tee structure and receiving its reflected or scattered echoes. The array controller is electrically connected to the ultrasonic transducer array, the transmit / receive switching circuit, and the data acquisition module. The array controller is used to generate a high-frequency excitation signal and control the transmit / receive switching circuit according to a preset array element selection order, so that the high-frequency excitation signal is applied to the selected transmit array element, and the echo signal received by the selected receive array element is transmitted to the data acquisition module.

4. The pipeline inspection robot based on composite resolution acoustic perception as described in claim 3, characterized in that: The pipeline inspection robot's inspection strategy for the target pipeline includes: first, using a low-frequency sensing component to perform a wide-area scan of the current section of the target pipeline within the visible range to identify the pipeline structure and the target to be inspected; then, based on the pipeline structure, performing path planning on the chassis to generate an inspection path that can traverse each target to be inspected within the current visible range, and generating an inspection list; then, driving the robot to approach each target to be inspected in turn, and performing ultrasonic imaging on it at the optimal observation distance. After completing the ultrasound imaging of all targets in the inspection list, repeat the same process to inspect the next section of the target pipeline until all pipelines in the target pipeline have been inspected.

5. The pipeline inspection robot based on composite resolution acoustic perception as described in claim 4, characterized in that: When the scanning results of the low-frequency sensing component indicate that the front end of the target pipe includes an impassable pipe structure, the controller drives the vehicle to approach and perform ultrasonic imaging to verify it; if the verification is successful, the inspection task is stopped, and then the robot is driven back along the original path and the abnormal information is reported. And / or, when the imaging results of the high-frequency sensing component reflect that the distance between the target and the probe exceeds the optimal observation distance, the controller drives the vehicle to adjust its position and then re-perform ultrasonic imaging.

6. The pipeline inspection robot based on composite resolution acoustic perception as described in claim 4, characterized in that: The pipeline structures identified in the scanning results of the low-frequency sensing component include empty pipes, blockages, aligned structures, tilted structures, branch pipe openings, and manhole openings; among them, blockages and manhole openings are impassable pipeline structures; the targets to be investigated include blockages, deposits, root intrusion, pipeline damage, tilted structures, and branch pipe openings.

7. The pipeline inspection robot based on composite resolution acoustic perception as described in claim 4, characterized in that: The data processing procedure implemented by the controller based on the echo signal from the low-frequency sensing component includes: (1) Bandpass filtering, denoising, and envelope extraction are performed on the original echo signal; (2) Identify the echo peak position according to the preset threshold and extract the effective signal contained therein; (3) Input each effective signal segment into a target classification model based on neural network pre-training to obtain the target classification result; (4) Calculate the distance of the target to the probe corresponding to each effective signal segment based on the sound speed and echo propagation time.

8. The pipeline inspection robot based on composite resolution acoustic perception as described in claim 1, characterized in that: The chassis is a motor-driven wheeled, tracked, support-wheeled, or other type of robot suitable for movement within a tube. The movement distance of the chassis is estimated by the motor speed collected by the encoder, thereby realizing the spatial positioning of the robot in the pipeline. And / or, the chassis also includes a gyroscope, which, based on the direction and acceleration information collected by the gyroscope, and in conjunction with the motor speed collected by the encoder, performs spatial positioning of the robot inside the pipeline.

9. The pipeline inspection robot based on composite resolution acoustic perception as described in claim 1, characterized in that: The controller includes a data storage module, which is used to store the signals collected by the low-frequency sensing component and the high-frequency sensing component, as well as the analysis results obtained after data processing. And / or, the probe also includes a camera with a supplementary light, and the controller is used to control the camera to start so as to acquire on-site images of each designated target; And / or, the power source is a rechargeable lithium battery.

10. An automatic pipeline inspection method based on composite resolution acoustic sensing, characterized in that, It employs a pipeline inspection robot based on composite resolution acoustic perception as described in any one of claims 1-9, and is used to perform segmented inspections of target pipelines based on known pipeline maps; It includes: A spatial map of the target pipeline is obtained in advance, an inspection path is generated based on the spatial map, and multiple wide-area scanning nodes are set on the inspection path according to the visibility range of the low-frequency sensing component. The robot starts from the starting point and arrives at the first wide-area scanning node. It uses low-frequency sensing components to perform a wide-area scan of the corresponding section of the target pipe within the visible range, and identifies the pipe structure and the target to be investigated contained therein. Based on the identified pipeline structure and the target to be inspected, the inspection path is analyzed for passability and optimized to generate an inspection path that can traverse all targets to be inspected within the current visible range, and an inspection list is generated. The robot is controlled to approach each target in turn and perform ultrasonic imaging at the optimal observation distance; the imaging results are dynamically binarized and clustered to obtain several feature clusters; three feature parameters of the feature clusters are calculated, including axial distance, axial length and non-zero angular proportion; the targets are classified and their structures are analyzed based on the feature parameters. After completing the ultrasonic imaging task for all targets in the inspection list, the controller drives the robot to the next wide-area scanning node, repeating the same process to complete the inspection task for the next section of the target pipeline, until the end of the inspection path is reached.

Images