Multi-sensor fusion pipeline interior detection device and system

Abstract

The invention discloses a multi-sensor fusion pipeline interior detection device and system, and relates to the technical field of industrial pipeline detection.The device comprises a mobile detection body, a sensor integration module and a detection mode switching chip, and the mobile detection body is used for driving and propelling along the interior of a pipeline; the sensor integration module is used for acquiring multi-modal sensing data; and the detection mode switching chip is integrated on the mobile detection body, establishes communication connection with the sensor integration module and is used for returning a first pipeline defect detection result. The technical problems that in traditional pipeline detection, due to the fact that a single sensor cannot adapt to a complex pipeline environment, the detection precision is insufficient, and internal defects of a pipeline cannot be effectively recognized are solved, and the purposes that multi-mode sensing data are obtained through multi-sensor fusion, the detection mode is intelligently adjusted in combination with a detection mode switching chip, and the detection efficiency is improved are achieved. The technical effects of high-precision identification and comprehensive evaluation of pipeline defects in a complex illumination environment are achieved.

Description

Technical Field

[0001] The present application relates to the technical field of industrial pipeline detection, and in particular to a multi-sensor fusion pipeline internal detection device and system. Background Art

[0002] Steel-lined pressure pipelines are widely used to transport corrosive media or fluids under extreme working conditions due to their excellent corrosion resistance and resistance to high temperature and high pressure. Such pipelines usually use polymer materials such as ceramics, enamel, and tetrafluoroethylene as the lining layer to extend their service life. However, the integrity of the lining layer is directly related to the safe operation of the pipeline. Minor coating cracks, pores, or uneven thickness may cause medium leakage or even explosion accidents. Currently, industry standards (such as TSGD7006-2020 and TSGD7005-2018) have put forward strict technical requirements for factory inspection and regular in-service inspection during the pipeline manufacturing stage. However, traditional detection methods often rely on manual or single sensor technology, resulting in detection efficiency and accuracy that are difficult to meet high-standard detection requirements in low-light, narrow or special pipeline structures. In addition, for the detection of lining coating materials such as ceramics, enamel, and tetrafluoroethylene, the limitations of traditional methods are even more obvious. It is impossible to simultaneously achieve multi-dimensional and multi-modal in-depth detection such as coating macro-inspection, thickness measurement, and electric spark detection. This makes comprehensive assessment of pipeline status and preventive maintenance difficult, thus affecting pipeline safe operation and lifecycle management. Summary of the Invention

[0003] This application provides a multi-sensor fusion pipeline internal detection device and system, which solves the technical problem that a single sensor in traditional pipeline detection cannot adapt to the complex pipeline environment, resulting in insufficient detection accuracy and inability to effectively identify internal defects in the pipeline. It achieves the technical effect of obtaining multi-modal perception data through multi-sensor fusion, and intelligently adjusting the detection mode by combining the detection mode switching chip, thereby realizing high-precision identification and comprehensive evaluation of pipeline defects in complex lighting environments.

[0004] The present application provides a multi-sensor fusion pipeline internal detection device, which includes: a mobile detection body, which is used to drive and advance along the interior of the pipeline; a sensor integration module, which includes an image sensor, a thickness sensor, an electric spark sensor, and at least one light sensor arranged on the mobile detection body, and is used to obtain multimodal sensing data, wherein the multimodal sensing data includes a surface image of the pipeline, coating thickness, electrical breakdown state, and light sensing data; a detection mode switching chip, which is integrated with the mobile detection body and establishes a communication connection with the sensor integration module; wherein, when the mobile detection body is driven and advanced along the interior of the pipeline, the detection mode switching chip determines whether the mobile detection body enters a preset low-light area based on the multimodal sensing data, and if so, activates a low-light detection mode to perform defect detection on the multimodal sensing data of the interior of the pipeline, and returns a first pipeline defect detection result.

[0005] Preferably, the step in which the detection mode switching chip determines whether the mobile detection body enters a preset low-light area based on the multimodal perception data includes: extracting the light intensity of the light perception data and the image frame brightness of the surface image; determining whether the light intensity of the light perception data is lower than a first preset threshold, and whether the image frame brightness of the surface image is lower than a second preset threshold; if the light intensity of the light perception data is lower than the first preset threshold, and the image frame brightness of the surface image is lower than the second preset threshold, entering the preset low-light area.

[0006] Preferably, the step in which the detection mode switching chip determines whether the mobile detection body enters a preset low-light area based on the multimodal sensing data also includes: if the illumination intensity of the illumination sensing data is lower than a first preset threshold, recording a first duration; if the image frame brightness of the surface image is lower than a second preset threshold, recording a second duration; if both the first duration and the second duration are greater than the preset time threshold, entering the preset low-light area.

[0007] Preferably, the detection mode switching chip switches to the low-light detection mode to perform defect detection on the multimodal perception data inside the pipeline, including: wherein the low-light detection mode includes low-light image acquisition configuration parameters and low-light image processing configuration parameters, the low-light image acquisition configuration parameter items include exposure time, gain value and frame rate, and the low-light image processing configuration parameter items include grayscale comparison value and Canny edge detection operator threshold; according to the low-light detection mode, low-light defect detection is performed on the surface image of the multimodal perception data, and image defect detection results are output; according to the coating thickness and the electrical breakdown state, thickness defect detection results and breakdown defect detection results are output; the first pipeline defect detection result is output based on the image defect detection result, thickness defect detection result and breakdown defect detection result.

[0008] Preferably, the values ​​of the low-light image acquisition configuration parameter items and the values ​​of the low-light image processing configuration parameter items are obtained through training of a defect detection effect evaluation model; the defect detection effect evaluation model is constructed based on a training data set of multiple groups of known defect sample images and trained until convergence.

[0009] Preferably, the detection mode switching chip further includes: if the mobile detection body enters the preset low-light area, determining whether the mobile detection body enters a preset high-light area based on the multimodal sensing data; if it enters the preset high-light area, exiting the low-light detection mode, restoring to the normal light detection mode, performing defect detection on the multimodal sensing data inside the pipeline, and returning a second pipeline defect detection result; recording a third pipeline defect detection result before the mobile detection body enters the preset low-light area; and obtaining a comprehensive pipeline defect detection result based on the first pipeline defect detection result, the second pipeline defect detection result, and the third pipeline defect detection result.

[0010] Preferably, the normal light detection mode includes normal light image acquisition configuration parameters and normal light image processing configuration parameters; wherein, the value of the normal light image acquisition configuration parameter item is different from the value of the low light image acquisition configuration parameter item, and the value of the normal light image processing configuration parameter item is different from the value of the low light image processing configuration parameter item.

[0011] Preferably, the mobile detection body is a crawler-type or wheel-type drive structure.

[0012] Preferably, the sensor integration module includes three light sensors, which are respectively arranged in front, on the left and on the right of the motion detection body.

[0013] The present application also provides a multi-sensor fusion pipeline internal detection system, which is used in a multi-sensor fusion pipeline internal detection device. The system includes: a memory for storing executable instructions; and a processor for executing the executable instructions stored in the memory.

[0014] This application proposes a multi-sensor fusion pipeline internal detection device and system. A mobile detection body is driven and advanced along the interior of a pipeline. A sensor integration module includes an image sensor, a thickness sensor, an electric spark sensor, and at least one light sensor, which are arranged on the mobile detection body and are used to obtain multimodal sensing data. The multimodal sensing data includes the surface image of the pipeline, coating thickness, electrical breakdown state, and light sensing data. A detection mode switching chip is integrated into the mobile detection body and establishes a communication connection with the sensor integration module. When the mobile detection body is driven and advanced along the interior of the pipeline, the detection mode switching chip determines whether the mobile detection body has entered a preset low-light area based on the multimodal sensing data. If it has entered the preset low-light area, the low-light detection mode is activated to perform defect detection on the multimodal sensing data of the pipeline interior, and a first pipeline defect detection result is returned. This solves the technical problem in traditional pipeline detection where a single sensor cannot adapt to complex pipeline environments, resulting in insufficient detection accuracy and inability to effectively identify internal pipeline defects. The technology achieves the technical effect of obtaining multimodal perception data through multi-sensor fusion, and intelligently adjusting the detection mode in combination with the detection mode switching chip, thereby realizing high-precision identification and comprehensive evaluation of pipeline defects in complex lighting environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention are briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in precise order. Instead, various steps may be processed in reverse order or simultaneously as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0016] Figure 1 A schematic structural diagram of a multi-sensor fusion pipeline internal detection device provided in an embodiment of the present application.

[0017] Figure 2 A schematic structural diagram of a multi-sensor fusion pipeline internal detection system provided in an embodiment of the present application.

[0018] Description of the accompanying drawings: motion detection body 11, sensor integrated module 12, detection mode switching chip 13, processor 21, memory 22, input device 23, output device 24. DETAILED DESCRIPTION

[0019] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0020] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0021] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict, and the terms “first\second” involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. The terms “including” and “having” and any variations are intended to cover non-exclusive inclusions. For example, a process, device, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or components that are not clearly listed or inherent to these processes, devices, products, or equipment. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only.

[0022] The present application provides a multi-sensor fusion pipeline internal detection device, such as Figure 1 As shown, the device includes:

[0023] The mobile detection body 11 is used to drive and propel along the inside of the pipeline.

[0024] Specifically, the mobile detection body 11 is the core structure of the entire internal pipeline detection device, responsible for autonomously or remotely controlling forward, backward or turning movements inside the pipeline to ensure that the detection device can cover the entire detection path. The mobile detection body 11 integrates a drive system, a power module, a positioning module, etc., so that it has good propulsion capabilities and stability, and can operate flexibly in complex or long-distance pipelines. In addition, the mobile detection body 11 also provides mechanical support and power interfaces for the sensor integration module 12 and the detection mode switching chip 13, ensuring the continuous operation and data transmission of the sensor integration module 12 and the detection mode switching chip 13, and is a key carrier for realizing the task of internal pipeline detection.

[0025] Further, the mobile detection body 11 is a tracked or wheeled driving structure.

[0026] Specifically, the mobile detection body 11 adopts a tracked or wheeled driving structure, aiming to improve its adaptability and passing ability in different types of pipelines. The tracked structure is suitable for pipelines with relatively smooth surfaces or certain obstacles, has good grip and stability, and can smoothly move in complex working conditions such as inclination, bending or water accumulation; while the wheeled structure has higher flexibility and movement speed, and is suitable for pipeline detection tasks with regular internal space and relatively smooth path. According to actual application requirements, different driving modes can be flexibly selected, and at the same time, the driving system is matched to realize precise control of the propulsion speed, direction and turning action, thereby ensuring the continuous and stable operation of the mobile detection body 11 inside the pipeline.

[0027] The sensor integration module 12 includes an image sensor, a thickness sensor, an electric spark sensor and at least one illumination sensor arranged on the mobile detection body 11, for acquiring multi-modal perception data, wherein the multi-modal perception data includes surface images of the pipeline, coating thickness, electric breakdown state and illumination perception data.

[0028] Specifically, the sensor integration module 12 is a key functional component installed on the mobile detection body 11, which integrates an image sensor, a thickness sensor, an electric spark sensor and at least one illumination sensor, aiming to synchronously collect multiple types of data during detection. The image sensor is used to acquire visual images of the internal surface of the pipeline, which is convenient for identifying surface defects such as cracks, corrosion and peeling; the thickness sensor is used to detect the thickness variation of the coating to evaluate the uniformity and integrity of the corrosion or protection layer; the electric spark sensor is used to detect whether there is electric breakdown in the non-metallic coating, which assists in judging whether the insulation performance is qualified; the illumination sensor senses the brightness of the internal environment of the pipeline in real time, which provides a basis for intelligent switching of subsequent detection modes. Through the cooperative work of these sensors, multi-modal perception data such as surface images, coating thickness, electric breakdown state and illumination intensity can be collected and fused, realizing comprehensive and stereoscopic perception of the internal state of the pipeline.

[0029] Further, the sensor integration module 12 includes three illumination sensors arranged at the front, left and right sides of the mobile detection body 11.

[0030] Specifically, three light sensors are arranged in the sensor integration module 12, which are respectively installed at the front, left and right positions of the mobile detection body 11, aiming to realize real-time monitoring of the light conditions in different directions inside the pipeline. Through this arrangement, the current detection environment can be fully perceived, and it can be determined whether there is an area with insufficient light, especially in complex pipeline environments with bends, weld shadows or structural obstructions, the local low-light area can be accurately identified. This arrangement not only enhances the adaptability of the detection device to environmental changes, but also provides accurate basis for subsequent detection mode switching related to light conditions (such as low-light detection mode), improving the intelligence and stability of the overall detection.

[0031] The detection mode switching chip 13 is integrated on the mobile detection body 11 and is in communication connection with the sensor integration module 12.

[0032] Specifically, the detection mode switching chip 13 is one of the control core elements integrated inside the mobile detection body 11, mainly responsible for data communication and state coordination with the sensor integration module 12. The detection mode switching chip 13 receives real-time multi-modal perception data collected from image sensors, thickness sensors, electric spark sensors and light sensors, and quickly analyzes and judges these data. Once a specific environmental condition change is detected (such as entering a low-light area), the detection mode switching chip 13 can automatically switch the current detection mode, for example, from normal mode to low-light detection mode, to optimize the effect of image acquisition and defect recognition. Through this real-time detection mode regulation mechanism, the detection mode switching chip 13 effectively improves the adaptability and detection accuracy of the detection device, and is the core control unit for realizing intelligent detection process.

[0033] When the mobile detection body 11 drives and advances along the inside of the pipeline, the detection mode switching chip 13 determines whether the mobile detection body 11 enters a preset low-light area according to the multi-modal perception data, and if it enters the preset low-light area, activates the low-light detection mode to detect defects in the multi-modal perception data of the inside of the pipeline, and returns the first pipeline defect detection result.

[0034] Specifically, when the mobile detection body 11 advances inside the pipeline, the detection mode switching chip 13 will continue to receive multimodal perception data from the sensor integration module 12, and analyze the lighting and image conditions reflected by these data in real time. If it is judged that the light intensity of the current environment is lower than the set threshold and the image brightness is insufficient, it is considered that the preset low-light area has been entered. At this time, it will automatically switch to the low-light detection mode, adjust the image acquisition parameters (such as exposure, gain, etc.), and apply an adaptive image processing algorithm to perform in-depth analysis of the data collected in the low-light environment, thereby realizing the identification of coating defects, thickness anomalies and electrical breakdown points. Finally, the pipeline defect detection results of the first stage are output to provide data support for subsequent comprehensive judgments.

[0035] Furthermore, the step of the detection mode switching chip 13 determining whether the motion detection body 11 enters a preset low-light area according to the multimodal sensing data includes:

[0036] Extract the illumination intensity of the illumination perception data and the image frame brightness of the surface image; determine whether the illumination intensity of the illumination perception data is lower than a first preset threshold, and whether the image frame brightness of the surface image is lower than a second preset threshold; if the illumination intensity of the illumination perception data is lower than the first preset threshold, and the image frame brightness of the surface image is lower than the second preset threshold, enter a preset low-light area.

[0037] Specifically, during the detection process, multiple light sensors deployed on the mobile detection body 11 first collect light perception data in real time and extract the current ambient light intensity from it. At the same time, a weighted calculation is performed on the RGB three-channel values ​​of the surface image collected by the image sensor, and the calculated pixel brightness is averaged to obtain the overall brightness value of the image frame, which serves as a key indicator for judging image clarity and visual quality. The weighted calculation formula is as follows: Y = 0.299 × R + 0.587 × G + 0.114 × B, where R is the red channel value, G is the green channel value, B is the blue channel value, and L is the pixel brightness value. Subsequently, the detection mode switching chip 13 compares the extracted light intensity with the image frame brightness to determine whether the light intensity is lower than a set first preset threshold (i.e., whether the ambient brightness is insufficient) and whether the image frame brightness is lower than a second preset threshold (i.e., whether the image is too dark). If both conditions are met, it means that the current ambient light inside the pipeline is insufficient and the image quality is low, indicating that there is a detection and recognition blind spot. At this point, the motion detection body 11 is determined to have entered a preset low-light area and triggers low-light detection mode, adjusting acquisition parameters and processing algorithms to ensure that image and data quality in subsequent detection processes meet recognition requirements. This process enables dynamic recognition of environmental changes and adaptive mode switching, a key mechanism for ensuring detection accuracy.

[0038] Furthermore, the step of the detection mode switching chip 13 determining whether the motion detection body 11 enters a preset low-light area according to the multimodal sensing data further includes:

[0039] If the illumination intensity of the illumination sensing data is lower than a first preset threshold, the first duration is recorded; if the image frame brightness of the surface image is lower than a second preset threshold, the second duration is recorded; if both the first duration and the second duration are greater than the preset time threshold, the preset low-light area is entered.

[0040] Specifically, when the illumination intensity of the illumination perception data is lower than a first preset threshold, the timing starts and the time during which the illumination intensity is lower than the threshold is recorded as the first duration. If the illumination intensity continues to be lower than the threshold for a certain period of time, it is considered that the area is insufficiently illuminated. When the image frame brightness of the surface image is lower than a second preset threshold, the timing also starts and the time during which the image frame brightness is lower than the threshold is recorded as the second duration. This step is used to ensure that the image quality remains poor for a period of time, and to avoid misjudgment of occasional low brightness of a single image. Only when both the first duration and the second duration exceed the preset time threshold will the current environment entered by the mobile detection body 11 be judged as a preset low-light area. This process effectively avoids misjudgment due to short-term ambient light fluctuations or instantaneous image brightness changes, ensuring that the low-light detection mode can be accurately activated, making the subsequent detection process more efficient and accurate.

[0041] Furthermore, the detection mode switching chip 13 switches to the low-light detection mode to perform defect detection on the multimodal sensing data inside the pipeline, including:

[0042] Among them, the low-light detection mode includes low-light image acquisition configuration parameters and low-light image processing configuration parameters, the low-light image acquisition configuration parameter items include exposure time, gain value and frame rate, and the low-light image processing configuration parameter items include grayscale comparison value and Canny edge detection operator threshold; low-light defect detection is performed on the surface image of the multimodal perception data according to the low-light detection mode, and an image defect detection result is output; according to the coating thickness and the electrical breakdown state, a thickness defect detection result and a breakdown defect detection result are output; the first pipeline defect detection result is output based on the image defect detection result, thickness defect detection result and breakdown defect detection result.

[0043] Specifically, in low-light detection mode, in order to adapt to insufficiently lit environments, the image acquisition and processing parameters are optimized and adjusted to ensure that clear and stable image data can be obtained even in poor lighting conditions. Specifically, in order to enhance the brightness of the image, the exposure time, gain value, and frame rate in the low-light image acquisition configuration parameter items are adjusted. For example, the exposure time is adjusted from the original 10ms to a longer time between 30ms and 50ms to capture more light and compensate for the dim image in low-light environments; the gain value is increased from 400 to 1600 to increase the brightness of the image, especially in low-light environments to enhance the clarity of image details and avoid images that are too dark to be recognized; the frame rate is reduced from 30fps to 10fps to slow down the image acquisition speed to improve the exposure quality of each frame and ensure image stability. In addition, the grayscale contrast value and Canny edge detection operator threshold in the low-light image processing configuration parameters are adjusted. For example, adjusting the grayscale contrast value to between 40 and 60 increases the grayscale contrast in the image, making subtle defects more prominent and facilitating subsequent defect detection and analysis. Setting the high threshold of the Canny edge detection operator to between 100 and 150 and the low threshold to between 30 and 50 better captures fine edges such as coating defects and cracks, ensuring accurate defect identification. The specific adjustment values ​​for these parameters can be obtained based on a pre-trained defect detection evaluation model. After completing the configuration of the low-light image acquisition and low-light image processing configuration parameters, the configured low-light detection mode is used to perform low-light defect detection on the surface image of the multimodal perception data. During this process, the grayscale value of each pixel in the image is adjusted based on the set grayscale contrast value, thereby enhancing the image contrast and making the details more visible. Subsequently, the Canny edge detector with a reset Canny edge detection operator threshold is applied to extract the edge information of the image by searching for gradient changes in the enhanced image, helping to identify small defects on the pipeline surface, such as cracks and corrosion points. The Canny edge detection results are then used to perform connectivity analysis on the edge areas in the image to determine whether the edge areas are defects (such as cracks, holes, and corrosion points). Based on the edge areas, regional expansion is then performed to merge close edge areas to form a larger defect area (for example, multiple small cracks are merged into a large crack area). Afterwards, the identified defect areas are marked, usually by highlighting the defect areas with rectangular frames, circular frames, or color filling, to obtain the image defect detection results.In addition to image defect detection, the coating thickness collected by the coating thickness sensor will be compared and analyzed with the standard coating thickness value. If the measured thickness is lower than the preset standard threshold, it will be judged that there is a coating thickness defect in the area, and a thickness defect detection result will be generated. For the electrical breakdown state collected by the electric spark sensor, a judgment will be made as to whether electrical breakdown has occurred. If the electrical breakdown state indicates the presence of a breakdown phenomenon (i.e., an abnormal signal), the electrical breakdown defect information of the area will be recorded, and a breakdown defect detection result will be generated. Finally, the image defect detection results, coating thickness defect detection results, and electrical breakdown defect detection results are aggregated into a set to output a complete first pipeline defect detection result, providing a comprehensive and accurate basis for the subsequent maintenance and repair of the pipeline. In summary, through this process, the detection efficiency and accuracy in low-light environments have been significantly improved, ensuring that comprehensive inspections inside the pipeline can be reliably completed even under complex environmental conditions.

[0044] Furthermore, the values ​​of the low-light image acquisition configuration parameter items and the values ​​of the low-light image processing configuration parameter items are obtained through training of a defect detection effect evaluation model; the defect detection effect evaluation model is constructed based on a training data set of multiple groups of known defect sample images and trained until convergence.

[0045] Specifically, the specific values ​​of exposure time, gain, and frame rate in the low-light image acquisition configuration parameter items, as well as the specific values ​​of grayscale comparison value and Canny edge detection operator threshold in the low-light image processing configuration parameter items, are obtained by inputting the currently monitored image into a pre-trained defect detection effect evaluation model. When constructing the defect detection effect evaluation model, multiple sets of known defect sample images are obtained from historical inspection logs and annotated using the corresponding historical lighting image acquisition configuration parameters and historical lighting image processing configuration parameters. Subsequently, the labeled sets of known defect sample images are divided into training, validation, and test sets. The convolutional neural network (CNN) model is trained using the training set. After training, the validation set is used for validation and evaluation of the model's performance. The CNN model consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. During training, the known defect sample images and their corresponding historical lighting image acquisition configuration parameters and image processing configuration parameters are input. The convolutional layers extract key image features, which are then reduced in dimension by the pooling layers. The fully connected layers fuse the extracted image features with the corresponding configuration parameters to output the most optimal illumination image acquisition and image processing configuration parameters. A mean squared error loss function is then used to calculate the loss between the predicted results and the actual configuration parameters. Backpropagation is used to calculate the gradient of the loss with respect to each layer's weights layer by layer. The Adam optimizer is then used to optimize the model parameters, adjusting the weights to minimize the loss function. The training process continues until the model converges. Finally, the model's performance is tested on a test set to evaluate its accuracy in predicting illumination image acquisition and image processing configurations. If the accuracy meets the target, the trained CNN model is output as the final defect detection performance evaluation model. Otherwise, hyperparameters such as the learning rate and batch size are adjusted to further improve the model's prediction performance.

[0046] Furthermore, the detection mode switching chip 13 further includes:

[0047] If the mobile detection body 11 enters the preset low-light area, determine whether the mobile detection body 11 enters the preset high-light area based on the multimodal sensing data; if it enters the preset high-light area, exit the low-light detection mode and restore to the normal light detection mode to perform defect detection on the multimodal sensing data inside the pipeline, and return a second pipeline defect detection result; record the third pipeline defect detection result before the mobile detection body 11 enters the preset low-light area; and obtain a comprehensive pipeline defect detection result based on the first pipeline defect detection result, the second pipeline defect detection result, and the third pipeline defect detection result.

[0048] Specifically, after the mobile detection body 11 enters a preset low-light area, it determines whether it has entered a preset high-light area based on the collected multimodal sensing data (including images, light perception, coating thickness, and electrical breakdown status). When a change in lighting conditions is detected, the light intensity and image frame brightness are analyzed and compared with first and second preset thresholds. If the light intensity and image frame brightness are not lower than the corresponding thresholds, it indicates that the mobile detection body 11 has entered a high-light area. At this point, the low-light detection mode is automatically exited and normal lighting detection mode is restored. After returning to normal lighting detection mode, defect detection continues based on the multimodal sensing data within the pipeline, outputting a second pipeline defect detection result. During this process, the defect detection effect evaluation model is used to readjust the image acquisition and image processing configuration parameters to adapt to the optimal lighting environment, thereby ensuring the accuracy of the detection results. Furthermore, to ensure the continuity and integrity of defect detection, a third pipeline defect detection result is recorded before the mobile detection body 11 enters the preset low-light area. This result reflects the detection status under normal lighting conditions. Finally, the first pipeline defect detection result, the second pipeline defect detection result, and the third pipeline defect detection result are added to a set in the order of the pipelines to obtain the comprehensive pipeline defect detection result, which provides a comprehensive basis for subsequent pipeline maintenance and repair.

[0049] Furthermore, the normal light detection mode includes normal light image acquisition configuration parameters and normal light image processing configuration parameters; wherein, the value of the normal light image acquisition configuration parameter item is different from the value of the low light image acquisition configuration parameter item, and the value of the normal light image processing configuration parameter item is different from the value of the low light image processing configuration parameter item.

[0050] Specifically, the normal light detection mode includes normal light image acquisition configuration parameters and normal light image processing configuration parameters. The acquisition method of these two types of parameters is the same as the aforementioned low light image acquisition configuration parameters and low light image processing configuration parameters, and both are obtained through the defect detection effect evaluation model. Compared with the low light mode, the exposure time under normal light will be set to a shorter value (such as reduced from 30-50ms in low light mode to 10-20ms). This is because the light is sufficient and a longer exposure time is not required to capture enough light; the gain value will also be set to a lower value (such as from 1600 to 400 or 800) to avoid overexposure of the image; at the same time, the frame rate can also be appropriately increased (such as from 10fps in low light mode to 30fps or higher), making the detection process more efficient. Similarly, the grayscale comparison value and the Canny edge detection operator threshold will also be set to values ​​that adapt to high light conditions, making the details in the image clearer and facilitating accurate detection of defects.

[0051] A multi-sensor fusion pipeline internal detection system according to an embodiment of the present invention includes: a processor 21, a memory 22, an input device 23, and an output device 24; the number of processors 21 in the system can be one or more. Figure 2 Taking a processor 21 as an example, the processor 21, memory 22, input device 23 and output device 24 in the system can be connected through a bus or other means. Figure 2 The bus connection is taken as an example.

[0052] Memory 22, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and components, such as the program instructions / components corresponding to a multi-sensor fusion pipeline interior detection device in embodiments of the present invention. Processor 21 executes the software programs, instructions, and components stored in memory 22 to perform various functional applications and data processing of the multi-sensor fusion pipeline interior detection device.

[0053] A multi-sensor fusion pipeline interior detection system provided in an embodiment of the present invention is used in a multi-sensor fusion pipeline interior detection device, and has corresponding functional components and beneficial effects of a multi-sensor fusion pipeline interior detection device.

[0054] Although this application makes various references to certain components in the system according to the embodiments of this application, any number of different components may be used and run on the user terminal and / or server, and the various units and components included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention.

[0055] The above specific embodiments do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. A multi-sensor fusion pipeline internal detection device, characterized in that: The device comprises: A mobile detection body is used to drive and propel along the inside of the pipeline; A sensor integration module, the sensor integration module including an image sensor, a thickness sensor, an electric spark sensor, and at least one light sensor disposed on the motion detection body, for acquiring multimodal sensing data, wherein the multimodal sensing data includes a surface image of the pipeline, coating thickness, electrical breakdown status, and light sensing data; A detection mode switching chip, which is integrated into the mobile detection body and establishes a communication connection with the sensor integration module; Among them, when the mobile detection body is driven along the inside of the pipeline, the detection mode switching chip determines whether the mobile detection body enters a preset low-light area based on the multimodal perception data. If it enters the preset low-light area, the low-light detection mode is activated to perform defect detection on the multimodal perception data inside the pipeline, and the first pipeline defect detection result is returned.

2. The pipeline internal detection device according to claim 1, characterized in that: The step of the detection mode switching chip determining whether the mobile detection body enters a preset low-light area according to the multimodal sensing data includes: extracting the illumination intensity of the illumination perception data and the image frame brightness of the surface image; determining whether the illumination intensity of the illumination perception data is lower than a first preset threshold, and whether the image frame brightness of the surface image is lower than a second preset threshold; If the illumination intensity of the illumination sensing data is lower than a first preset threshold, and the image frame brightness of the surface image is lower than a second preset threshold, the system enters a preset low-light area.

3. The pipeline internal detection device according to claim 2, characterized in that: The step of the detection mode switching chip determining whether the mobile detection body enters a preset low-light area according to the multimodal sensing data further includes: If the illumination intensity of the illumination sensing data is lower than a first preset threshold, recording a first duration; If the image frame brightness of the surface image is lower than a second preset threshold, recording a second duration; If both the first duration and the second duration are greater than a preset time threshold, the system enters a preset low-light area.

4. The pipeline internal detection device according to claim 1, characterized in that: The detection mode switching chip switches to the low-light detection mode to perform defect detection on the multimodal sensing data inside the pipeline. include: The low-light detection mode includes low-light image acquisition configuration parameters and low-light image processing configuration parameters. The low-light image acquisition configuration parameter items include exposure time, gain value and frame rate. The low-light image processing configuration parameter items include grayscale comparison value and Canny edge detection operator threshold. performing low-light defect detection on the surface image of the multimodal perception data according to the low-light detection mode, and outputting an image defect detection result; Outputting a thickness defect detection result and a breakdown defect detection result according to the coating thickness and the electrical breakdown state; The first pipeline defect detection result is output based on the image defect detection result, the thickness defect detection result, and the breakdown defect detection result.

5. The pipeline internal detection device according to claim 4, characterized in that: The values ​​of the low-light image acquisition configuration parameter items and the values ​​of the low-light image processing configuration parameter items are obtained through defect detection effect evaluation model training; The defect detection effect evaluation model is constructed based on a plurality of groups of known defect sample images to construct a training data set and is trained until convergence is achieved.

6. The pipeline internal detection device according to claim 1, characterized in that: The detection mode switching chip further includes: If the motion detection body enters the preset low-light area, determining whether the motion detection body enters the preset high-light area based on the multimodal sensing data; If entering the preset high-light area, exiting the low-light detection mode, reverting to the normal-light detection mode, performing defect detection on the multimodal sensing data inside the pipeline, and returning a second pipeline defect detection result; Recording a third pipeline defect detection result before the mobile detection body enters the preset low-light area; A comprehensive pipeline defect detection result is obtained according to the first pipeline defect detection result, the second pipeline defect detection result, and the third pipeline defect detection result.

7. The pipeline internal detection device according to claim 6, characterized in that: The normal light detection mode includes normal light image acquisition configuration parameters and normal light image processing configuration parameters; The value of the normal-light image acquisition configuration parameter item is different from the value of the low-light image acquisition configuration parameter item, and the value of the normal-light image processing configuration parameter item is different from the value of the low-light image processing configuration parameter item.

8. The pipeline internal detection device according to claim 1, characterized in that: The mobile detection body is a crawler-type or wheel-type drive structure.

9. The pipeline internal detection device according to claim 1, characterized in that: The sensor integration module includes three light sensors, which are respectively arranged in front, on the left and on the right of the motion detection body.

10. A multi-sensor fusion pipeline internal detection system, characterized in that: The system is used in a multi-sensor fusion pipeline internal detection device according to any one of claims 1 to 9, comprising: a memory for storing executable instructions; A processor is configured to execute the executable instructions stored in the memory.

Images