Underwater oil pipeline leakage detection method based on acoustic imaging and image recognition

Abstract

The invention relates to the technical field of underwater detection, and discloses an underwater oil pipeline leakage detection method based on acoustic imaging and image recognition, and the method comprises the steps: firstly, employing wide-beam acoustic imaging to recognize a candidate region with a plume form; then switching to a high-coherence narrow beam mode, detecting the phase of the reference point echo after passing through the candidate area, and calculating a phase jitter quantitative index; performing cooperative judgment in combination with morphological characteristics and phase jitter indexes, and confirming a leakage event; and finally, performing three-dimensional scanning on the confirmed leakage area, constructing a phase disturbance intensity spatial distribution diagram, and determining accurate three-dimensional coordinates of a leakage source according to a peak point of the phase disturbance intensity spatial distribution diagram. Through bimodal combination of the morphological characteristics of the target and the physical characteristic of sound wave phase jitter caused by leakage turbulence, the method can effectively distinguish real leakage from acoustic false targets such as fish schools and suspended silt, significantly reduces the detection false alarm rate, and achieves high-precision three-dimensional positioning of a leakage source.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of underwater detection, in particular to an underwater oil pipeline leakage detection method based on acoustic imaging and image recognition. BACKGROUND

[0002] Underwater pipelines are the key infrastructure for large-scale and long-distance oil and gas transportation. These pipelines have been serving in complex marine environments for a long time, facing multiple risks such as corrosion, geological activities, and third-party damage, and there is a possibility of leakage. Oil and gas leakage not only causes huge economic losses and energy waste, but also causes long-term and serious damage to the marine ecosystem. Therefore, developing a technology that can quickly and accurately detect and locate the leakage point of underwater pipelines is of great significance to energy security and marine environment protection.

[0003] In the prior art, various methods for detecting underwater pipeline leakage have been developed. For example, the visual detection method based on optical camera, through the underwater vehicle carrying a high-definition camera to patrol along the pipeline. However, the effectiveness of this method depends heavily on the cleanliness of the water body, and in the common marine environment with high turbidity, its detection distance and recognition ability will decrease sharply. In addition, some oil and gas leakage (such as natural gas) forms a plume in water, which is transparent or semi-transparent, which also brings challenges to optical recognition.

[0004] Another method is acoustic detection method. Among them, the passive acoustic method listens to the characteristic acoustic signals generated by the high-pressure fluid spouting when the pipeline leaks by laying hydrophones. The limitation of this method is that the interference of marine environment background noise (such as ship navigation, marine biological activity, drilling platform operation, etc.) is very serious, and the weak leakage signal is easy to be overwhelmed by noise. At the same time, it is difficult to accurately locate the leakage point by passive listening.

[0005] To overcome the above-mentioned defects, active acoustic method, especially imaging technology based on forward-looking sonar, is applied to leakage detection. This technology generates underwater acoustic images by actively emitting sound waves and receiving echo signals of the target. When the pipeline leaks, the oil and gas mixture spouted forms a plume composed of a large number of micro-bubbles or oil droplets in water, which has a significant acoustic impedance difference with the surrounding water, and thus appears as an abnormal target with a specific shape on the acoustic image.

[0006] Although the imaging method based on active sonar is superior to optical method in detection distance and anti-turbidity ability, it still faces a core technical problem in practical application: high false alarm rate. The fundamental reason is that this method mainly relies on the morphological characteristics of the target for judgment. In complex marine environment, there are many acoustic pseudo-targets similar in shape to the leakage plume, such as dense fish shoals, suspended silt clouds formed by water flow stirring, submarine hydrothermal vents, etc. The traditional acoustic imaging system cannot effectively distinguish these targets with similar physical causes, resulting in frequent misjudgment of pseudo-targets as real leaks, thus causing unnecessary emergency response and reducing the reliability and efficiency of the detection system.

[0007] In addition, even if the traditional active acoustic method correctly identifies the real leakage plume, accurate positioning of the leakage source is another technical challenge. The leakage plume will drift and diffuse under the action of environmental factors such as ocean currents, and it appears as a large area with irregular shape on the acoustic image. The area with the strongest acoustic reflection on the image usually does not accurately correspond to the actual leakage point on the pipeline, which brings difficulties to subsequent maintenance operations. SUMMARY

[0008] In view of the shortcomings of the prior art, the present application provides an underwater oil pipeline leakage detection method based on acoustic imaging and image recognition, which solves the problem that the existing active acoustic detection method only relies on the morphological characteristics of the leakage plume for judgment, thus cannot effectively distinguish acoustic pseudo-targets with different physical causes (such as fish shoals and suspended silt clouds), resulting in high false alarm rate of detection, and difficulty in accurately positioning the leakage source in three dimensions through acoustic image due to the drift and diffusion of the plume.

[0009] To achieve the above purpose, the present application is realized by the following technical scheme: an underwater oil pipeline leakage detection method based on acoustic imaging and image recognition, which specifically comprises the following steps: S1, an acoustic image of the environment around the underwater pipeline is obtained by using a wide-beam imaging mode, and an image recognition algorithm is used to process the acoustic image to output a candidate area suspected of leakage in shape and morphological characteristics of the candidate area.

[0010] In one embodiment, the detection system is mounted on an underwater vehicle and cruises along the pipeline. The system works in wide-beam imaging mode, emits wide-angle acoustic pulses, and obtains acoustic images of the environment around the pipeline in a large range. The image recognition algorithm analyzes the real-time acquired acoustic image and identifies targets with plume or cloud shape characteristics. The morphological characteristics can be morphological confidence. When the morphological confidence of a target is higher than a first preset threshold, the target is determined as a candidate area, and its bounding box information is recorded.

[0011] In an optional embodiment, the acoustic image is also pre-processed by a fuzzy weighted average filter before being identified by the image recognition algorithm. The pre-processing procedure is as follows: a sliding window is used to traverse all pixels in the acoustic image, and each pixel in the window is modeled by a fuzzy set; the membership degree of each pixel in the window is determined by iterative optimization calculation; the weight of each pixel in the window is calculated according to the final membership degree; finally, the gray value of each pixel in the window is weighted and averaged according to the weight, and the filtered central pixel value is obtained to suppress random noise in the acoustic image.

[0012] S2, after identifying the candidate region, switching the detection system to a high-coherence narrow-beam detection mode, emitting a detection sound beam to the candidate region, and receiving an echo signal from a preset reference point.

[0013] Once the candidate region is identified in the S1 step, the underwater vehicle will approach the region, and the working mode of the sonar system will be switched from the wide-beam imaging mode to the high-coherence narrow-beam detection mode. In this mode, the system emits a narrow-beam detection sound beam with high phase stability to the candidate region. The detection sound beam passes through the candidate region and is reflected by a preset reference point with stable acoustic reflection characteristics. In an embodiment, the preset reference point is the outer wall of the pipeline behind the candidate region. The system receives the echo signal returned by the reference point.

[0014] To ensure the accuracy of the phase measurement, the detection sound beam is preferably a linear frequency modulation signal, and its mathematical expression is: ; wherein, is the transmitted signal; is the signal amplitude; is a rectangular window function, which is equal to 1 in the pulse duration and 0 at other times, and it defines the transmission interval of the signal; is the pulse duration; is the time variable; is the center frequency of the signal; is the frequency modulation rate of the linear frequency modulation, which is equal to the ratio of the signal bandwidth to the pulse duration It is because of the quadratic phase term that the instantaneous frequency of the signal changes linearly within the pulse duration; is the cosine function.

[0015] S3, extracting the phase information of the echo signal and calculating the phase jitter quantization index of the phase information.

[0016] The physical principle of this step is that the oil-water mixture formed by real leakage is an acoustically inhomogeneous and time-varying medium. When the sound wave passes through this medium, the propagation distance of the sound wave will randomly fluctuate, and this fluctuation is finally reflected in the phase jitter of the echo signal. By quantifying this phase jitter, the physical properties of the medium can be judged.

[0017] The specific implementation of this step is to use a digital phase-locked loop to track the echo signal received in step S2 and extract its instantaneous phase. Then, the phase jitter signal representing the random fluctuation is separated from the instantaneous phase. Then, the variance of the phase jitter signal in a preset observation time window is calculated, and the calculated variance value is taken as the phase jitter quantization index. The calculation formula of the variance is: ; wherein, is the phase jitter quantization index, i.e., the variance of the phase jitter; is the length of the observation time window; is the phase jitter signal varying with time; is the mean value of the phase jitter signal in the observation time window; represents the square of the difference between the instantaneous value of the signal at any time and its mean value; is the definite integral operation, which represents the accumulation of the instantaneous fluctuation power (i.e., the square term) of the signal in the entire observation time .

[0018] S4, the morphological features obtained in step S1 are fused with the phase jitter quantization index obtained in step S3 to cooperatively judge and finally confirm the leakage event.

[0019] This step establishes a dual-mode cooperative judgment criterion. The criterion is that when the morphological confidence of the candidate region obtained in step S1 is higher than a first preset threshold, and the phase jitter quantization index of the candidate region calculated in step S3 is higher than a second preset threshold, the system determines that the candidate region is a real leakage event. The second preset threshold is calibrated by measuring the baseline level of the phase jitter in a background environment without leakage. This cooperative judgment mechanism can effectively distinguish between real leakage targets caused by oil-water mixture and pseudo targets caused by suspended sediment or biological groups, etc., which are only morphologically similar but have no phase disturbance characteristics.

[0020] In a further embodiment, after confirming the leakage event in step S4, the method further comprises the step of locating the position of the leakage source: S5, control the high-coherence narrow-beam to grid scan the candidate region confirmed as the leakage event, measure the phase jitter quantization index of multiple spatial points in the scanning grid, and construct the spatial distribution of the phase disturbance intensity. S6, determining the peak point in the spatial distribution of the phase disturbance intensity as the leakage source position.

[0021] The principle of the positioning step is that the leakage source is the root cause of the most severe disturbance of the medium physical characteristics, and therefore the phase jitter quantization index of the corresponding spatial point is also the largest. By performing three-dimensional grid scanning on the confirmed leakage area and calculating the phase jitter quantization index of each grid point, a three-dimensional spatial distribution map of the phase disturbance intensity can be constructed. The leakage source position is determined by finding the global maximum point of the distribution map: ; wherein, is the spatial coordinate in the scanning area; is the phase jitter quantization index corresponding to the coordinate; is the phase jitter quantization index; is a mathematical operator.

[0022] The present application provides an underwater oil pipeline leakage detection method based on acoustic imaging and image recognition. It has the following beneficial effects: 1. The present application uses the technical means of synergistic judgment of the morphological features and phase jitter quantization index of the candidate area, which requires that the detected target not only appears to be leaking in the acoustic image in terms of morphology, but also can cause significant disturbance to the phase of sound wave propagation in terms of physical characteristics, so as to effectively distinguish between real oil-water mixture caused leakage and pseudo-targets formed by suspended silt, biological groups, etc., and improve the accuracy and reliability of the leakage event detection.

[0023] 2. The present application realizes direct tracing of the leakage source by performing grid scanning on the candidate area after confirming the leakage event, and determining the leakage source position by the peak point of the phase disturbance intensity spatial distribution, which directly corresponds to the physical damage point of the pipeline, avoids the positioning error caused by the drift of plume due to ocean current, and obtains a high-precision positioning result.

[0024] 3. The present application uses a wide-beam imaging mode for large-scale preliminary screening and a high-coherence narrow-beam detection mode for close inspection, which combines the work flow of large-scale preliminary screening and close inspection, ensures the efficiency of pipeline inspection in a large area, and concentrates high-precision physical verification resources on a small number of identified candidate areas, taking into account the detection efficiency and detection accuracy, so that the overall detection method is more practical and robust in actual application. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 is the overall architecture block diagram of the underwater pipeline leakage detection system of the present application; Figure 2The overall flowchart of the underwater pipeline leakage detection method of the present application; Figure 3 The schematic diagram of the physical property verification process in steps S2 and S3 of the present application; Figure 4 The logic block diagram of the dual-mode collaborative judgment in step S4 of the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the present application specification. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0027] Refer to the drawings Figure 1 , Figure 1 The overall architecture block diagram of the underwater pipeline leakage detection system according to an embodiment of the present application is shown in FIG. 1. The present application provides an underwater oil pipeline leakage detection system 100, which can include: an underwater vehicle 10, a dual-mode forward-looking sonar system 20, a signal processing unit 30, and a main control and decision unit 40.

[0028] The underwater vehicle 10 is used as a carrying platform of the entire detection system, and is used to carry the dual-mode forward-looking sonar system 20, the signal processing unit 30, and the main control and decision unit 40. The underwater vehicle 10 receives navigation instructions from the main control and decision unit 40, and performs cruising along a preset pipeline path or performs approaching, hovering, and scanning in a specified area.

[0029] The dual-mode forward-looking sonar system 20 is installed at the front end of the underwater vehicle 10, and is used to emit acoustic pulses and receive echo signals. The dual-mode forward-looking sonar system 20 has two working modes that can be switched by the main control and decision unit 40: a wide-beam imaging mode and a high-coherence narrow-beam detection mode. In the wide-beam imaging mode, the system emits wide-angle acoustic pulses to cover a large area; in the high-coherence narrow-beam detection mode, the system emits narrow-beam acoustic pulses with high phase stability to detect a specific area.

[0030] The signal processing unit 30 is electrically connected with the dual-mode forward-looking sonar system 20, and is used to receive and preliminarily process the original echo signals collected by the dual-mode forward-looking sonar system 20. In the wide-beam imaging mode, the signal processing unit 30 is used to construct an acoustic image according to the time-of-flight and intensity information of the echo signals. In the high-coherence narrow-beam detection mode, the signal processing unit 30 is used to perform pulse compression processing on the echo signals and extract the phase information thereof. The signal processing unit 30 sends the processed acoustic image data or phase data to the main control and decision unit 40.

[0031] The master and decision unit 40 is communicatively coupled to the underwater vehicle 10, the dual-mode forward-looking sonar system 20, and the signal processing unit 30. The master and decision unit 40 is configured to receive acoustic images and phase information outputted by the signal processing unit 30, run image recognition algorithms to process the acoustic images to identify candidate regions and output morphological features, control the dual-mode forward-looking sonar system 20 to switch between the two modes according to pre-set conditions, run a collaborative decision logic to fuse the morphological features and the phase jitter quantization index to confirm a leak event, run a fine positioning algorithm after confirming the leak event, and send navigation instructions to the underwater vehicle 10 to control its motion attitude.

[0032] In the overall workflow, the system 100 first controls the underwater vehicle 10 to cruise along the pipeline, and the dual-mode forward-looking sonar system 20 works in the wide-beam imaging mode to continuously generate acoustic images. The master and decision unit 40 analyzes the acoustic images in real time to identify candidate regions. Then, the master and decision unit 40 controls the underwater vehicle 10 to approach the candidate region and instructs the dual-mode forward-looking sonar system 20 to switch to the high-coherence narrow-beam detection mode. The signal processing unit 30 and the master and decision unit 40 collaboratively calculate the phase jitter quantization index and finally confirm the leak event. After confirming the leak, a scanning positioning procedure can be further performed to determine the precise location of the leak source.

[0033] The functions of the detection system can be implemented by a series of functional modules executed in the master and decision unit 40 or the signal processing unit 30. The functional modules include an acoustic imaging module 210, an image processing module 220, a mode control module 230, an acoustic detection module 240, a signal processing module 250, a collaborative decision module 260, and a fine positioning module 270.

[0034] The acoustic imaging module 210 is configured to control the dual-mode forward-looking sonar system 20 to emit acoustic pulses with a wide angle in the wide-beam imaging mode and receive raw echo data from a large range of water areas, and pass the raw echo data to the signal processing unit 30 for constructing acoustic images.

[0035] The image processing module 220 is configured to receive and process the acoustic images constructed by the signal processing unit 30. The image processing module 220 executes image recognition algorithms to identify targets with pre-set morphologies from the acoustic images and output bounding box information of one or more candidate regions and morphological features corresponding to each candidate region. In an embodiment, the morphological features are morphological confidence.

[0036] The mode control module 230 is configured to generate and send a mode switching instruction. When the image processing module 220 identifies the candidate region, the mode control module 230 sends an instruction to the dual-mode forward-looking sonar system 20 to switch from the wide-beam imaging mode to the high-coherence narrow-beam detection mode.

[0037] The acoustic detection module 240 is configured to control the dual-mode forward-looking sonar system 20 to emit a detection beam to the candidate region determined by the image processing module 220 and receive the echo signal reflected by the preset reference point after the system switches to the high-coherence narrow-beam detection mode.

[0038] The signal processing module 250 is configured to process the echo signal received by the acoustic detection module 240. The signal processing module 250 executes one or more digital signal processing algorithms, extracts the phase information from the echo signal, and calculates the phase jitter quantization index. The signal processing module 250 outputs the calculated phase jitter quantization index to the collaborative decision module 260.

[0039] The collaborative decision module 260 is configured to receive the morphological features from the image processing module 220 and the phase jitter quantization index from the signal processing module 250. The collaborative decision module 260 fuses the two input information according to a preset collaborative judgment criterion and finally outputs a decision result for confirming whether the candidate region is a real leakage event.

[0040] The precise positioning module 270 is activated after the collaborative decision module 260 confirms the leakage event. The precise positioning module 270 controls the acoustic detection module 240 and the signal processing module 250 to perform grid scanning and measurement on the confirmed leakage region, obtains a spatial distribution of phase disturbance intensity, and calculates the three-dimensional coordinate position of the leakage source according to the spatial distribution.

[0041] The detection method provided by the present application is realized through a multi-stage and multi-modal process. The process starts with a wide-range and rapid inspection, gradually focuses on a small-range accurate verification and positioning, and the complete workflow includes a preliminary screening stage, a physical property verification stage, and a leakage source positioning stage.

[0042] In the preliminary screening stage, the system first performs step S1. The underwater vehicle 10 carrying the entire detection system cruises along the pipeline, at which time the dual-mode forward-looking sonar system 20 works in the wide-beam imaging mode. In this mode, the sonar system transmits acoustic waves to a large range of water area and receives echoes, and the signal processing unit 30 continuously constructs and outputs acoustic images around the underwater pipeline. The image processing module 220 in the main control and decision unit 40 processes the acoustic images in real time, runs the image recognition algorithm, identifies the targets with plume or cloud shape in the acoustic images, and outputs one or more candidate regions and the corresponding morphological features. The function of this stage is to complete the rapid preliminary screening of the candidate regions suspected of leakage in morphology.

[0043] In the physical property verification stage, after the candidate region is identified, the system then performs steps S2, S3 and S4. The main control and decision unit 40 controls the underwater vehicle 10 to approach the candidate region determined in step S1, and at the same time, the mode control module 230 sends instructions to switch the dual-mode forward-looking sonar system 20 from the wide-beam imaging mode to the high-coherence narrow-beam detection mode. In this mode, the acoustic detection module 240 controls the sonar to transmit a detection beam to the candidate region, and receives the echo signals reflected by the preset reference points such as the outer wall of the pipeline. The signal processing module 250 extracts the phase information of the echo signals and calculates the phase jitter quantization index. Finally, the cooperative decision module 260 fuses the morphological features obtained in step S1 and the phase jitter quantization index obtained in step S3, and according to the preset cooperative judgment criterion, finally confirms the authenticity of the candidate region and determines whether it is a real leakage event.

[0044] In the leakage source positioning stage, after the leakage event is confirmed in step S4, the system can further perform steps S5 and S6. The precise positioning module 270 is activated and controls the high-coherence narrow-beam to perform three-dimensional grid scanning on the candidate region confirmed as a leakage event. During the scanning process, the measurement process of step S3 is repeated for each grid point to obtain the phase jitter quantization index of the space point, thereby constructing a three-dimensional spatial distribution of phase disturbance intensity. Finally, the precise positioning module 270 searches for the peak point in the spatial distribution of phase disturbance intensity, and outputs the three-dimensional coordinates of the peak point as the final positioning result of the leakage source.

[0045] Referring to the accompanying drawings Figure 2 , Figure 2 is the overall flowchart of the underwater pipeline leakage detection method according to the present application. In a specific embodiment, step S1 is described in detail. The execution subject of this step is the dual-mode forward-looking sonar system 20, the signal processing unit 30 and the image processing module 220 in the main control and decision unit 40.

[0046] At the beginning of a detection task, the master and decision unit 40 controls the dual-mode forward-looking sonar system 20 to operate in a wide-beam imaging mode. In this mode, the transducer array of the dual-mode forward-looking sonar system 20 is excited to emit a fan-shaped acoustic beam with a wide opening angle in the horizontal direction and a narrow opening angle in the vertical direction. This acoustic beam is capable of covering a large area in front of the underwater vehicle 10 at one time, thus achieving a fast scan of the pipeline and its surrounding environment.

[0047] The emitted acoustic pulses propagate in the water, interact with the underwater oil pipeline, the seabed, and the suspended targets in the water body, and generate echo signals. The transducer array of the dual-mode forward-looking sonar system 20 receives these echo signals and converts the acoustic pressure signals into analog electrical signals, which are then transmitted to the signal processing unit 30.

[0048] The signal processing unit 30 amplifies, filters, and digitizes the received multi-channel analog electrical signals to obtain digitized echo data. Then, the signal processing unit 30 performs digital beamforming processing on the digitized echo data to determine the directions of arrival of different echo signals. At the same time, the signal processing unit 30 records the round-trip time of each echo signal. The target distance is calculated as follows: ; wherein, is the speed of sound in water, is a time variable.

[0049] The signal processing unit 30 constructs a two-dimensional acoustic image based on the calculated target distance , direction of arrival , and echo intensity I of each echo. In this acoustic image, the position of a pixel is determined by the target distance and the direction of arrival , and the gray value of the pixel is proportional to the echo intensity I.

[0050] As the underwater vehicle 10 continuously moves along the pipeline path, the above-mentioned acoustic wave emission, echo reception, and image construction processes are repeatedly performed at a preset frame rate, thereby generating a continuous acoustic image sequence. The signal processing unit 30 transmits this acoustic image sequence to the image processing module 220 in the master and decision unit 40 in real time for subsequent recognition and processing.

[0051] In an optional embodiment, to suppress speckle noise in the acoustic image and enhance the signal-to-noise ratio of potential targets, the image processing module 220 also performs a fuzzy weighted average filtering preprocessing step before the acoustic image is sent to the image recognition algorithm.

[0052] The pre-processing step traverses each pixel in the input acoustic image by a sliding window of size M x M. For a window centered at the current pixel, all pixels within the window are considered as a local pixel set, and the subsequent filtering operation is performed on this set.

[0053] Within the sliding window, N pixels (N = M x M) are modeled for fuzzification, which aims to partition the pixel set into C classes (e.g., C = 2, corresponding to target signal class and background noise class, respectively). This process is achieved by minimizing an objective function, which is specifically formulated as: ; wherein, C is the preset number of classes; N is the total number of pixels within the window; is the gray value of the th pixel within the window; is the cluster center of the th class; is the membership degree of the th pixel belonging to the th class, satisfying ; is a fuzzy weighting index greater than 1, used to control the degree of fuzziness; denotes the squared Euclidean distance between the feature vector of the pixel and the feature vector of the cluster center ; is denoted as the summation over all data samples (pixels); is denoted as the summation over all cluster classes.

[0054] To obtain the optimal membership degree that minimizes the objective function , the image processing module 220 performs iterative optimization calculation on the membership degree of the pixels within the window. In each iteration, the membership degree is updated according to the cluster center of the previous round, and then the new cluster center is calculated according to the updated membership degree . The update formulas for the membership degree and the cluster center are respectively: ; wherein, is the distance between the pixel and the current target cluster center ; is the distance between the same pixel and all cluster centers ; This ratio measures the distance from a pixel to the target cluster center. The distance relative to its distance to another cluster center The magnitude of the distance; It is a key index term, directly influenced by the fuzzy weighted index. Control; It is a constant greater than 1; For a fixed pixel and a target category We need to iterate through all the categories (from arrive ),calculate The value of this item, then this Add up all the calculation results.

[0055] ; in, It is a weighted sum; Clustering of all pixels The membership weights are summed, which serves as a normalization process.

[0056] The above iterative process continues until the objective function is between two iterations. The change in the number of iterations is less than a preset convergence threshold, or the preset maximum number of iterations is reached.

[0057] After the iteration is complete, each pixel within the window is obtained. For target signal category final membership degree Based on the final optimized membership degrees, the weights of each pixel within the window are calculated. In one embodiment, weight The calculation method is as follows: ; in, It is the degree of membership in a specific category; It represents the first in the image. 1 pixel; It is a specific category index; It is a fuzzy weighted index; It is a pixel The non-normalized contribution; It is the total contribution and the normalization factor.

[0058] Finally, based on the weights grayscale values ​​of all pixels within the window Perform a weighted average, and use the resulting weighted average as the filtered output value of the center pixel of the sliding window. The formula is: ; wherein, is the normalized centroid calculation weight; is the gray value of the pixel; is the weighted sum operator.

[0059] By traversing the entire acoustic image with a sliding window, a frame of acoustic image preprocessed by fuzzy weighted average filtering and noise suppressed can be obtained for subsequent image recognition processing.

[0060] After obtaining the acoustic image or the preprocessed acoustic image, the image is transmitted to the image processing module 220 integrated in the main control and decision unit 40 for execution of candidate region generation.

[0061] The image processing module 220 is internally deployed with a pre-trained image recognition algorithm, which in one embodiment is a target detection model based on a convolutional neural network (CNN). The model has been trained on a dataset containing a large number of labeled underwater acoustic images, including samples of oil and gas plumes formed by real leaks and samples of various non-leakage targets (such as suspended silt, biological groups, structures, etc.).

[0062] The image recognition algorithm receives a frame of acoustic image as input and analyzes the image content to identify targets that conform to the morphological characteristics of a leak plume.

[0063] The output of the algorithm is a set of detection results, each of which corresponds to an identified target.

[0064] For each identified target, the image recognition algorithm outputs its bounding box (Bounding Box) and morphological confidence. The bounding box is used to indicate the location and range of the target in the acoustic image, which can be represented by a four-tuple .

[0065] wherein, are the horizontal and vertical coordinates of the top-left corner of the bounding box in the image coordinate system, respectively; are the width and height of the bounding box, respectively.

[0066] The morphological confidence is the quantitative output of the image recognition algorithm evaluating the similarity of the target morphology to the morphology of the pre-trained leak sample, with a value range of. A higher value indicates that the morphology of the target is highly consistent with the morphological characteristics of the leak plume.

[0067] After obtaining all the identified targets and their corresponding morphological confidence, the image processing module 220 determines the final candidate regions according to a preset determination condition. The determination condition is: ; wherein, is a preset first preset threshold value. When the morphological confidence of a target is higher than the first preset threshold value , the target is confirmed as a candidate region. The first preset threshold value is calibrated according to the detection sensitivity and false alarm rate requirements of the actual application scene.

[0068] Finally, the image processing module 220 outputs the boundary box information of all candidate regions satisfying the above condition for guiding the subsequent physical property verification stage. If no target satisfying the condition is detected in a frame of acoustic image, the system continues to perform wide-beam imaging scanning.

[0069] In step S1, when the image processing module 220 determines one or more candidate regions, the mode control module 230 in the main control and decision unit 40 is triggered to start step S2.

[0070] The mode control module 230 sends a mode switching instruction to the dual-mode forward-looking sonar system 20. After receiving the instruction, the dual-mode forward-looking sonar system 20 switches its working mode from the wide-beam imaging mode to the high-coherence narrow-beam detection mode. In this mode, the transducer array of the sonar is configured by the phase shifter weighting to form a narrow beam with high directivity, so as to concentrate acoustic energy to a specific direction where the candidate region provided by the image processing module 220 is located.

[0071] After switching to the high-coherence narrow-beam detection mode, the acoustic detection module 240 controls the signal generator of the dual-mode forward-looking sonar system 20 to generate and transmit a detection beam to the candidate region. In a specific embodiment, in order to obtain high signal-to-noise ratio and high-precision phase information required for subsequent processing, the transmitted detection beam is a linear frequency modulation (LFM) signal.

[0072] The complex expression of the linear frequency modulation signal is: ; wherein, is the amplitude of the signal; is a rectangular window function, which is defined as 1 when , and 0 otherwise; is the pulse duration of the signal; is the time variable; is the initial frequency of the signal; is the frequency modulation slope, which has a value of , where is the bandwidth of the signal; is the imaginary unit; is essentially a rotating unit vector; is the total phase function.

[0073] The linear frequency modulation signal has a large time-bandwidth product, and after pulse compression processing by the signal processing unit 30, a narrow main lobe width and high processing gain can be obtained, which provides a technical basis for accurately extracting phase information from the echo signal subsequently.

[0074] Referring to the accompanying drawings Figure 3 , Figure 3 is a schematic diagram of the physical property verification process in steps S2 and S3 of the present application. After the acoustic detection module 240 controls the dual-mode forward-looking sonar system 20 to emit a detection beam, the detection beam propagates along the predetermined direction, passes through the candidate region determined by step S1, and irradiates a preset reference point located distally from the candidate region.

[0075] In a specific embodiment, the preset reference point is the outer wall of the oil pipeline to be measured. The outer wall of the pipeline is selected as the reference point because it has a significant acoustic impedance difference relative to the water body, which can form a high-intensity and time-stable acoustic reflection interface.

[0076] After the detection beam is reflected via the preset reference point, an echo signal is formed. The echo signal passes through the candidate region again and is received by the receiving transducer array of the dual-mode forward-looking sonar system 20. The receiving channel of the sonar converts the captured acoustic pressure signal into an analog electrical signal.

[0077] During the round trip through the candidate region, the phase of the echo signal is modulated by the random fluctuations of the acoustic refractive index of the medium (i.e., the substance within the candidate region) along the propagation path. Therefore, the received echo signal contains information about the physical properties of the medium in the candidate region. Its signal model can be represented as: ; where: is the total attenuation coefficient of the signal during propagation; is the transmitted signal component after double-path propagation delay ; is the phase modulation term introduced by the medium; is the random phase modulation term accumulated by the acoustic wave during the round trip through the medium in the candidate region; is the sum of environmental noise and internal system noise.

[0078] The received analog echo signal containing phase modulation information is transmitted to the signal processing unit 30 for subsequent phase information extraction and quantization index calculation in step S3.

[0079] Upon receiving the echo signal from the dual-mode forward-looking sonar system 20, the signal processing unit 30 or the signal processing module 250 integrated therein starts to perform step S3, i.e. extracting the phase information of the echo signal and calculating the phase jitter quantization index.

[0080] Firstly, the signal processing unit 30 performs pulse compression processing on the received digital echo signal. This processing is achieved by convolving the echo signal with a local conjugate matched filter of the transmitted linear frequency modulation signal. The output of the pulse compression is a signal with narrow pulse width and high signal-to-noise ratio, whose peak position accurately corresponds to the distance of the preset reference point (the outer wall of the pipeline).

[0081] Then, the system sets a time gate at the peak of the pulse-compressed output signal, and within this time gate, a digital phase-locked loop (PLL) is used to track the echo signal. The digital phase-locked loop locks and tracks the phase of the echo signal reflected by the preset reference point. Within a continuous observation time window, the digital phase-locked loop outputs a time series representing the instantaneous phase of the echo signal.

[0082] The instantaneous phase contains multiple components, which can be represented by the following equation: ; wherein: is a fixed phase constant related to the initial state of the propagation path and system parameters; is the Doppler shift caused by the relative motion between the underwater vehicle and the pipeline; is the linear phase drift term caused by the Doppler shift; is the phase jitter signal caused by the random fluctuations of the refractive index of the medium (such as the oil-water mixture) when the sound wave passes through the candidate region, which is the target signal to be extracted; is a time variable.

[0083] In order to separate the phase jitter signal from the instantaneous phase , the signal processing module 250 performs detrending processing on the time series within a preset observation time window . In an embodiment, this processing is achieved by least squares method on​​ A first order polynomial fit is performed to obtain a linear trend term The trend term is then subtracted from the original instantaneous phase to obtain a phase jitter signal: ; After obtaining the time series of phase jitter signal , the signal processing module 250 calculates the variance of the series within the same observation time window , and takes the variance as the final phase jitter quantization indicator . For discrete sampling points, the calculation formula is: ; Where: is the total number of sampling points within the observation time window ; is the discrete summation operator; is the phase jitter value at the th sampling time; is the arithmetic mean of all in the window, since the value after detrending is close to 0; is the variance contribution of a single sample; : indicates the specific time of the th sampling point on the time series.

[0084] After calculation, the signal processing module 250 outputs the obtained phase jitter quantization indicator to the collaborative decision module 260 in the main control and decision unit 40, which is used to perform the collaborative judgment of step S4.

[0085] Referring to the attached Figure 4 , Figure 4 is the logic block diagram of the dual-mode collaborative judgment in step S4 of the present application. After the signal processing module 250 calculates and outputs the phase jitter quantization indicator , the system starts to perform step S4, which is performed by the collaborative decision module 260 in the main control and decision unit 40.

[0086] The collaborative decision module 260 is used to receive two independent input data: The morphological confidence for the current candidate region generated by the image processing module 220 in step S1. .

[0087] The phase jitter quantization indicator for the same candidate region generated by the signal processing module 250 in step S3. .

[0088] The function of the collaborative decision module 260 is to fuse the two input data based on a pre-defined dual-modal collaborative decision criterion, and output a confirmation result about whether the candidate region is a real leakage event.

[0089] In a specific embodiment, the dual-modal collaborative decision criterion is defined as a set of logical conditions that must be satisfied simultaneously. Only when the candidate region is similar to the leakage plume in morphological features and can cause significant phase jitter of the acoustic signal in physical properties, the candidate region is finally confirmed as a real leakage event.

[0090] The decision criterion contains the following two independent conditions: Condition 1: Morphological confidence of the candidate region must be greater than a first pre-set threshold .

[0091] ; wherein: is a quantitative value output by the image recognition algorithm in step S1, representing the degree of similarity between the morphology of the candidate region and the morphology of the leakage plume. is a pre-calibrated threshold value used to preliminarily screen out targets highly suspected of leakage in morphology from the image. Meeting this condition is a prerequisite for the target to become a candidate region.

[0092] Condition 2: Phase jitter quantitative index of the candidate region must be greater than a second pre-set threshold .

[0093] ; wherein: is a quantitative value calculated in step S3, representing the degree of disturbance of the phase experienced by the sound wave when passing through the candidate region. is a pre-calibrated threshold value, the value of which is determined by measuring the background water body without leakage multiple times to obtain the statistical distribution of the background phase noise. The setting of ensures that only when the phase jitter is significantly higher than the normal background noise level, the condition is met.

[0094] The final decision logic of the collaborative decision module 260 is: only when the above condition 1 and condition 2 are met simultaneously, the collaborative decision module 260 outputs the signal of "leakage event confirmation". If either condition is not met, the signal of "non-leakage event" is output, and the system is instructed to abandon the current candidate region and return to the wide-range cruise scanning state of step S1. The confirmed leakage event information will be used to activate the precise positioning module 270 of step S5.

[0095] The application of the dual-mode collaborative judgment criterion enables the detection method of the present application to effectively distinguish a real leakage event from an acoustic false target. The acoustic false target refers to an object similar in shape to a leakage plume but not formed by a pipeline leakage in terms of physical causes, such as a fish school or a suspended silt cloud.

[0096] For an acoustic false target (for example, a suspended silt cloud), in the acoustic image generated by the wide-beam imaging mode, since it also presents a cloud or plume shape without a fixed boundary, the image processing module 220 outputs a higher morphological confidence when the image recognition algorithm is executed in step S1 . When this value satisfies , the acoustic false target is mistakenly identified as a candidate region, thereby triggering the subsequent physical property verification stage.

[0097] However, although the shape of the acoustic false target meets condition one, its physical composition is essentially different from a real oil and gas leakage. The suspended silt cloud is a mixture of solid particles and water, and the acoustic refractive index of the internal medium is different from that of pure water, but the internal structure is relatively stable, or the turbulence intensity is much lower than the intense turbulence caused by high-pressure oil and gas eruption.

[0098] Therefore, when the system switches to the high-coherence narrow-beam detection mode and emits a detection sound beam to the false target region, the phase of the sound wave is weakly and slowly modulated in the round trip through the region. This results in a very small amplitude of the phase jitter signal extracted by the signal processing module 250 in step S3.

[0099] Correspondingly, the value of the calculated phase jitter quantization index is also low, which does not meet the preset judgment condition. That is: ; wherein is a second preset threshold for distinguishing significant phase jitter from background noise.

[0100] Finally, when the collaborative decision module 260 performs collaborative judgment, since the acoustic false target does not meet condition two ( ), the logical AND operation result of the dual-mode collaborative judgment criterion is false. The collaborative decision module 260 accordingly outputs the decision result of "non-leakage event", thereby excluding the acoustic false target from the candidate region and avoiding false alarms. This process ensures that only the target that meets both the morphological feature and the physical property (significant phase jitter) is finally confirmed as a real leakage.

[0101] In step S4, after the collaborative decision module 260 confirms the leakage event, the accurate positioning module 270 in the master and decision unit 40 is activated to start step S5. The function of this step is to construct a phase disturbance intensity spatial distribution that can represent the location of the leakage source.

[0102] The accurate positioning module 270 first defines a three-dimensional scanning space in the coordinate system of the underwater vehicle 10 according to the candidate area boundary box determined in step S1 and the leakage event information confirmed in step S4. Then, the three-dimensional scanning space is discretized into a three-dimensional grid composed of a plurality of preset discrete grid points .

[0103] Wherein: are the coordinates of the grid point in the three-dimensional coordinate system, respectively; are the indices of the grid point in the three dimensions, respectively.

[0104] Next, the accurate positioning module 270 starts to scan and measure the three-dimensional grid point by point. For each grid point , the accurate positioning module 270 performs the following operations: First, the acoustic detection module 240 is controlled to accurately direct the transmission beam of the dual-mode forward-looking sonar system 20, so that the central axis of the high-coherence narrow beam passes through the current grid point and finally irradiates the preset reference point (i.e. the outer wall of the pipeline) located behind it.

[0105] Then, the system receives and processes the echo signal reflected by the reference point. The accurate positioning module 270 calls the signal processing module 250 to repeatedly perform the complete processing procedure in step S3 on the echo signal, i.e. pulse compression, phase extraction, detrending, and calculation of a phase jitter quantization index. The index here is one-to-one corresponding to the grid point, denoted as .

[0106] The accurate positioning module 270 traverses all the preset grid points in the three-dimensional grid to perform the above scanning and measuring process. After completing the measurement of all points, a three-dimensional spatial distribution of phase disturbance intensity is obtained. The distribution is a data set containing the coordinates of each spatial grid point and the phase jitter quantization index measured at the point. Its mathematical form can be expressed as: ; Wherein: is the three-dimensional spatial distribution of phase disturbance intensity; is a three-dimensional grid point; is the phase jitter quantization index measured when the detection beam path passes through the grid point; ​is a logical symbol; is the dimension index of the grid; represents all legal indices The set of combinations.

[0107] The meaning of this entire phrase is: this set contains all the (coordinate, phase perturbation intensity value) pairs corresponding to each point in the predetermined three-dimensional grid, without omission.

[0108] This three-dimensional spatial distribution map is stored in the storage unit of the precise positioning module 270 and serves as direct input data for performing step S6, i.e., determining the precise coordinates of the final leakage source.

[0109] After constructing the three-dimensional spatial distribution map of phase perturbation intensity in step S5 , the system then proceeds to step S6. This step is continued by the precise positioning module 270 in the main control and decision unit 40, which functions to determine the precise three-dimensional coordinates of the leakage source from the three-dimensional spatial distribution map.

[0110] The physical basis of this step is that the source of the pipeline leakage is the location where the high-pressure oil and gas is injected into the surrounding water body, and thus is the spatial point with the maximum turbulent intensity and the most drastic change in medium physical properties. This drastic turbulence directly leads to the maximum random fluctuation of the acoustic refractive index on the acoustic propagation path, thereby producing the strongest phase perturbation. Therefore, in the three-dimensional spatial distribution map of phase perturbation intensity , the spatial position corresponding to the peak point is the location of the leakage source.

[0111] The precise positioning module 270 receives the three-dimensional spatial distribution map generated in step S5 as input data. Subsequently, the precise positioning module 270 performs a peak search operation on this data set, which traverses all the discrete grid points in the distribution map and compares the phase perturbation quantization index corresponding to each grid point.

[0112] The three-dimensional coordinates of the leakage source are determined by finding the grid point that maximizes the phase perturbation quantization index . This determination process can be represented by the following formula: ; where: is the three-dimensional spatial coordinates of the finally determined leakage source; is a function that returns the parameter (in this case, the grid point ) that maximizes the subsequent expression; is a discrete point in the three-dimensional scanning grid; is a phase jitter quantification index measured on a path centered on a grid point ; is a complete three-dimensional spatial distribution map containing all grid points and their corresponding values generated by step S5.

[0113] After the peak search operation is completed, the precise positioning module 270 outputs the obtained three-dimensional coordinates of the leakage source as the final result of the entire detection process. This coordinate information can be recorded, stored, or sent to the upper computer for display.

Claims

1. A method for detecting leaks in underwater oil pipelines based on acoustic imaging and image recognition, characterized in that, Includes the following steps: S1. Acoustic images around the underwater pipe are acquired using a wide-beam imaging mode, and the acoustic images are processed using an image recognition algorithm to output candidate areas that are suspected of leaking in terms of morphology and the morphological features of the candidate areas. S2. After identifying the candidate region, the detection system is switched to the high coherence narrow beam detection mode, the detection sound beam is emitted towards the candidate region, and the echo signal from the preset reference point is received. S3. Extract the phase information of the echo signal and calculate the phase jitter quantization index of the phase information; S4. The morphological features obtained in step S1 and the phase jitter quantification index obtained in step S3 are used for collaborative judgment to finally confirm the leakage event.

2. The underwater oil pipeline leak detection method based on acoustic imaging and image recognition according to claim 1, characterized in that, In step S1, the image recognition algorithm outputs the bounding box and morphological confidence score of the candidate region, and the condition for determining the candidate region is that the morphological confidence score is higher than a first preset threshold.

3. The underwater oil pipeline leak detection method based on acoustic imaging and image recognition according to claim 1, characterized in that, In step S2, the preset reference point is the outer wall of the pipe.

4. The underwater oil pipeline leak detection method based on acoustic imaging and image recognition according to claim 1, characterized in that, In step S2, the emitted detection beam is a linear frequency modulated signal.

5. The underwater oil pipeline leakage detection method based on acoustic imaging and image recognition according to claim 1, characterized in that, The specific implementation method for extracting the phase information of the echo signal and calculating the phase jitter quantization index in step S3 is as follows: the echo signal is tracked using a phase-locked loop to obtain the instantaneous phase, and the phase jitter signal is separated from the instantaneous phase; the variance of the phase jitter signal within an observation time window is calculated, and the variance is used as the phase jitter quantization index.

6. The underwater oil pipeline leak detection method based on acoustic imaging and image recognition according to claim 1, characterized in that, The specific criteria for collaborative judgment in step S4 are as follows: when the morphological confidence level corresponding to the morphological feature is higher than the first preset threshold and the phase jitter quantization index is higher than the second preset threshold, the candidate region is confirmed as a leakage event.

7. The underwater oil pipeline leak detection method based on acoustic imaging and image recognition according to claim 1, characterized in that, After confirming the leak event in step S4, the method further includes the step of locating the source of the leak: S5. Control the high coherence narrow beam to perform gridded scanning on the candidate region that has been confirmed as a leakage event, measure the phase jitter quantization index of multiple spatial points in the scanning grid, and construct the spatial distribution of phase disturbance intensity; S6. Determine the peak point in the spatial distribution of the phase disturbance intensity as the location of the leakage source.

8. The underwater oil pipeline leak detection method based on acoustic imaging and image recognition according to claim 1, characterized in that, The function of the wide-beam imaging mode is to perform a rapid cruise scan over a large area to complete the initial screening of the candidate areas. The function of the high-coherence narrow-beam detection mode is to perform close-range physical characteristic verification of the identified candidate areas to confirm the authenticity of the leakage event.

9. The underwater oil pipeline leak detection method based on acoustic imaging and image recognition according to claim 1, characterized in that, In step S4, the collaborative judgment can distinguish between the oil-water mixture formed by a real leak and the false target formed by suspended sediment and biomass.

10. The underwater oil pipeline leak detection method based on acoustic imaging and image recognition according to claim 1, characterized in that, In step S1, before using the image recognition algorithm for recognition, the method further includes a step of performing a fuzzy weighted average filtering preprocessing on the acoustic image. The fuzzy weighted average filtering preprocessing step specifically includes: The acoustic image is traversed through a sliding window, and the pixels are modeled with blurring within the window; The membership degree of pixels within the window is iteratively optimized and calculated. Calculate the weight of each pixel within the window based on the final optimized membership degree; The grayscale values ​​of all pixels within the window are weighted according to the weights, and the filtered pixel values ​​are output.

Images