Pipeline defect detection optimization method, electronic device and storage medium

Abstract

The present application relates to the technical field of data processing, in particular to a pipeline defect detection optimization method, an electronic device and a storage medium, the method comprising: according to the detection videos respectively shot by a given pipeline defect detection device in a preset pipeline at each preset flow rate, obtaining detection information of a plurality of preset defect types corresponding to each preset flow rate, combining actual defect information corresponding to the given pipeline, constructing a target fitting curve of a first defect detection score and flow rate, determining a target point in the curve, extending the flow rate of the target point to determine a flow rate interval, introducing jitter data to obtain a second defect detection score, and finally determining a target flow rate and target jitter data as the adjustment reference of the pipeline defect detection device; the present application can accurately determine the reasonable adjustment threshold of the detection device, ensure a high detection accuracy during pipeline defect detection, and reduce the operation complexity and adjustment cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a pipeline defect detection optimization method, an electronic device and a storage medium. BACKGROUND

[0002] As a key infrastructure for transporting water and other fluids, pipelines are essential in residential communities, office buildings and other living places, so the structural integrity and safety of pipelines are of great importance. Due to corrosion, aging, external damage and other reasons, various defects often occur on the inner wall of the pipeline, so it is crucial to regularly detect defects in the pipeline to prevent leaks and ensure production safety.

[0003] Currently, visual-based pipeline defect detection devices, such as camera systems carried by pipeline robots, are gradually applied to such detection tasks. Such devices travel inside the pipeline along with the fluid, and their travel speed is close to the fluid speed. By shooting video streams and performing image analysis by a backend system, various defect types can be identified and classified. However, in actual detection operations, the running state of the detection device, especially its travel speed in the pipeline fluid and the image jitter caused by fluid disturbance and equipment vibration, will affect the final defect detection accuracy. Moreover, when the flow speed is too low, although clearer images can be obtained, it will lead to low detection efficiency, especially for long-distance pipelines, which will consume a lot of time and energy. SUMMARY

[0004] To solve the above technical problems, the present application provides a pipeline defect detection optimization method, an electronic device and a storage medium, which can accurately determine the reasonable adjustment threshold of the detection device, while ensuring high pipeline defect detection accuracy, and reducing the operation complexity and adjustment cost.

[0005] According to a first aspect of the present application, a pipeline defect detection optimization method is provided, comprising the following steps:

[0006] According to the detection videos respectively shot by a given pipeline defect detection device at each preset flow speed corresponding to the water flow in a preset pipeline, detection information of a plurality of preset defect types corresponding to each preset flow speed is obtained respectively.

[0007] According to the detection information of a plurality of preset defect types and the actual defect information corresponding to the given pipeline, a target fitting curve of defect detection score and flow speed is constructed.

[0008] A target point with a defect detection score not lower than a first preset score threshold is determined from the target fitting curve, and the flow speed of the target point is extended to determine a flow speed interval.

[0009] Based on n groups of preset jitter data, n groups of defect detection information corresponding to each preset flow rate in the flow rate interval are collected respectively, and a second defect detection score corresponding to each group of defect detection information is obtained, to obtain a corresponding relationship between each group of jitter data and the second defect detection score at each preset flow rate.

[0010] According to the corresponding relationship, a target flow rate and target jitter data are determined from a plurality of preset flow rates and a plurality of preset jitter data corresponding to a second defect detection score not lower than a second preset score threshold, to optimize and adjust the corresponding module of the given pipeline defect detection device by taking the target flow rate and the target jitter data as the adjustment threshold.

[0011] According to a second aspect of the present application, a non-transitory computer readable storage medium is provided, the storage medium storing at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by a processor to implement the above-mentioned pipeline defect detection optimization method.

[0012] According to a third aspect of the present application, an electronic device is provided, comprising a processor and the above-mentioned non-transitory computer readable storage medium.

[0013] The present application has at least the following beneficial effects:

[0014] The present application provides a pipeline defect detection optimization method, which first obtains detection information of a plurality of preset defect types corresponding to each preset flow rate according to detection videos respectively photographed by a given pipeline defect detection device at each preset flow rate in a preset pipeline, calculates a first defect detection score in combination with actual defect information corresponding to the given pipeline, and constructs a target fitting curve of the first defect detection score and flow rate. By constructing the target fitting curve, the trend of the detection score changing with the flow rate can be obtained, which provides a favorable basis for screening a suitable flow rate threshold. Then, a target point in the curve is determined, and the flow rate of the target point is extended to determine a flow rate interval, which ensures higher detection accuracy and makes the limitation condition of subsequent optimization and adjustment of the detection device more relaxed. Second defect detection scores are obtained by introducing jitter data, which makes the obtained defect detection scores closer to the actual detection situation, provides more accurate threshold adjustment reference, and finally determines the target flow rate and the target jitter data as the adjustment reference of the pipeline defect detection device. By accurately determining the reasonable adjustment threshold of the detection device, the detection accuracy is ensured while the operation complexity and adjustment cost are reduced during pipeline defect detection. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to make the technical solutions in the embodiments of the present application clearer, the accompanying drawings needed in the embodiment description will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present application, and all other embodiments obtained by those skilled in the art without creative efforts based on these accompanying drawings also belong to the protection scope of the present application.

[0016] Figure 1 The flowchart of the pipeline defect detection optimization method provided by the embodiments of the present application. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative efforts also belong to the protection scope of the present application.

[0018] The embodiments of the present application provide a pipeline defect detection optimization method, as shown in Figure 1 The method comprises the following steps:

[0019] S1, according to the detection videos respectively photographed by a given pipeline defect detection device at each preset flow rate corresponding to the water flow in a preset pipeline, respectively acquiring detection information of a plurality of preset defect types corresponding to each preset flow rate; it can be understood that the detection information of the preset defect type includes the label of the preset defect type and the defect feature text corresponding to the preset defect type.

[0020] Preferably, the detection videos are identified by a pre-trained image defect recognition model to obtain the detection information of the plurality of preset defect types; for example, the image defect recognition model can be a multi-modal large model. Alternatively, human detection can be performed according to requirements.

[0021] Specifically, the plurality of preset defect types include but are not limited to internal wall attachments, corrosion, holes, obstacles, cracks, wear and interface defects.

[0022] S2, according to the detection information of the plurality of preset defect types and the actual defect information corresponding to the given pipeline, constructing a target fitting curve of the first defect detection score and the flow rate.

[0023] In one specific embodiment, the target fitting curve is constructed by the following steps:

[0024] S201, according to the detection information of the plurality of preset defect types and the actual defect information corresponding to the given pipeline, calculating a first recognition proportion of the preset defect type at each preset flow rate, and a first recognition score corresponding to each recognized preset defect type.

[0025] Specifically, the first recognition proportion refers to a proportion of a number of the recognized preset defect types to a total number of preset defect types at a corresponding preset flow rate.

[0026] Preferably, the first recognition score corresponding to each of the preset defect types is obtained by the following steps:

[0027] S2011, for any preset defect type, based on a plurality of preset feature labels, the detection information of the quantifiable features and the detection information of the non-quantifiable features of the preset defect type are determined from the detection information of the preset defect type; it can be understood that the plurality of preset feature labels include quantifiable features and non-quantifiable features, i.e. quantifiable features and non-quantifiable features, for example, when the preset defect type is a crack, the corresponding quantifiable features include length, width, and the non-quantifiable features include texture pattern, edge feature, and direction, etc. The texture pattern can be linear, net-like, and tree-like, and the direction can be longitudinal, circumferential, etc. When the preset defect type is corrosion, the corresponding quantifiable features include corrosion area and maximum pit depth, and the non-quantifiable features include topography type, clothing shape, and distribution form, etc. which will not be repeated here.

[0028] S2012, for the quantifiable features, the detection information of the quantifiable features is calculated with the actual defect information of the corresponding preset defect type of the given pipeline, and the defect similarity corresponding to the quantifiable features is calculated. For example, the ratio of the detected defect length to the actual length is determined as the defect similarity.

[0029] S2013, for the non-quantifiable features, the semantic similarity corresponding to the non-quantifiable features is calculated according to the detection information of the non-quantifiable features and the actual defect information of the corresponding preset defect type of the given pipeline; it can be understood as the similarity of the labels in the text.

[0030] S2014, the defect similarity corresponding to the quantifiable features and the semantic similarity corresponding to the non-quantifiable features are input into the pre-trained machine learning model, and the first recognition score of each preset defect type is obtained.

[0031] In the above, when calculating the first recognition score of the preset defect type, multiple dimensions of multiple defect features of the preset defect type are comprehensively analyzed, both the quantifiable features and the non-quantifiable features are considered, and the finally obtained first recognition score is more reasonable and accurate.

[0032] S202, the first recognition proportion corresponding to any preset flow rate and the plurality of first recognition scores are weighted and calculated to obtain the first defect detection score corresponding to any preset flow rate.

[0033] Preferably, the weights of the plurality of first recognition scores are the same and are smaller than the weight of the first recognition proportion.

[0034] S203, constructing a target fitting curve of the first defect detection score and the flow rate according to the first defect detection score corresponding to each preset flow rate.

[0035] The first defect detection score corresponding to each preset flow rate is accurately obtained, the detection score of the detection device under different flow rate states is analyzed, the detection accuracy of the detection device under different flow rate states is reflected, the change trend of the detection score with the flow rate is obtained by constructing the target fitting curve, and a favorable basis is provided for screening a suitable flow rate threshold.

[0036] S3, determining a target point with a defect detection score not lower than a first preset score threshold from the target fitting curve, and extending the flow rate of the target point to determine a flow rate interval.

[0037] Specifically, the flow rate of the target point is extended to determine the flow rate interval, including the following steps:

[0038] S301, when the number of preset flow rates not greater than the flow rate corresponding to the target point is greater than a preset number threshold, the target point is extended in the direction of decreasing flow rate, and the flow rate interval is determined according to the maximum and minimum values in the several preset flow rates corresponding to the flow rate from 0 to the new target point.

[0039] S302, when the number of preset flow rates not greater than the flow rate corresponding to the target point is not greater than the preset number threshold, the flow rate interval is determined according to the maximum and minimum values in the several preset flow rates not greater than the flow rate corresponding to the target point.

[0040] In the specific implementation, because the greater the flow rate, the more blurred the corresponding detection video, and the lower the defect detection score, when the number of preset flow rates meeting the condition is large, the target point can be moved forward, the number of flow rate samples is reduced, and the preset flow rate is generally reduced to adapt to the condition that the subsequent joining of the jitter still enables the shooting of high video clarity, and when the number of preset flow rates meeting the condition is small, the flow rate interval is directly determined, the rationality of the flow rate sample selection is realized, the selected flow rate interval has high video clarity and detection accuracy, and the limitation condition for subsequent optimization and adjustment of the detection device is more relaxed, and the adjustment cost is reduced.

[0041] S4, based on n groups of preset jitter data, n groups of defect detection information corresponding to each preset flow rate in the flow rate interval are collected, and a second defect detection score corresponding to each group of defect detection information is obtained, and a corresponding relationship between each group of jitter data and the second defect detection score under each preset flow rate is obtained.

[0042] Preferably, the preset jitter data comprises a preset jitter frequency and a preset jitter amplitude. The preset jitter amplitude can be a fixed empirical value.

[0043] Specifically, the second defect detection score corresponding to each group of defect detection information is obtained through the following steps:

[0044] S401, for any preset flow rate in the flow rate interval, according to the n groups of defect detection information corresponding to the preset flow rate, and combining the actual defect information corresponding to the given pipeline, the second recognition proportion of each group of defect detection information under the preset defect type is calculated, and the second recognition score corresponding to each preset defect type is recognized. The second recognition score is obtained in the same way as the first recognition score, and will not be repeated here.

[0045] Specifically, the second recognition proportion refers to the proportion of the number of recognized preset defect types to the total number of preset defect types under any group of defect detection information corresponding to any preset flow rate.

[0046] S402, the second recognition proportion and the second recognition score corresponding to each group of defect detection information are respectively weighted and calculated, and the second defect detection score corresponding to each group of defect detection information is obtained.

[0047] Preferably, the weight of the second recognition score is the same, and is less than the weight of the second recognition proportion.

[0048] The above, after determining the flow rate interval with high detection accuracy, the jitter condition often appears in actual operation is added on the basis of several flow rates in the flow rate interval, and the defect detection score is calculated again. By considering the two dimensions of flow rate and detection device jitter, the obtained defect detection score is more reliable and more consistent with the actual detection condition, and a more accurate threshold adjustment reference is provided.

[0049] S5, according to the corresponding relationship, the target flow rate and the target jitter data are determined from the several preset flow rates and the several preset jitter data corresponding to the second defect detection score not less than the second preset score threshold, so as to take the target flow rate and the target jitter data as the adjustment threshold to optimize and adjust the corresponding module of the given pipeline defect detection device.

[0050] In one specific embodiment, the target flow rate and the target jitter data are determined through the following steps:

[0051] S501, according to the second defect detection score corresponding to each preset flow rate under n groups of preset jitter data, a detection score matrix of m*n is constructed; wherein m is the number of preset flow rates in the flow rate interval.

[0052] S502, determining, from the detection score matrix, a plurality of second defect detection score corresponding preset flow rate and preset jitter data association pairs not lower than a second preset score threshold.

[0053] S503, traversing the plurality of association pairs, and obtaining, from the association pairs, a quantity of preset flow rates corresponding to each preset jitter data.

[0054] S504, determining, from a plurality of preset jitter data, a maximum jitter data corresponding to a quantity of preset flow rates not less than a preset flow rate quantity proportion.

[0055] S505, traversing the plurality of association pairs, and obtaining, from the association pairs, a maximum preset flow rate corresponding to the target jitter data as the target flow rate.

[0056] The above, first obtain a plurality of association pairs satisfying the second preset score threshold and analyze the plurality of association pairs, then find out the jitter data that can satisfy the second preset score threshold under most flow rates, and take the jitter data as the adjustment threshold to match a plurality of preset flow rates satisfying the second preset score threshold. Matching a plurality of preset flow rates is conducive to screening out a larger flow rate, and taking the larger flow rate as the threshold can reduce the adjustment times while ensuring a higher detection accuracy, thereby reducing the operation complexity and adjustment cost.

[0057] Further, the target flow rate and the target jitter data are taken as the adjustment threshold to optimize and adjust the corresponding module of the given pipeline defect detection device, including the following steps:

[0058] S510, when the real-time flow rate and the real-time jitter data of the given pipeline defect detection device are both not greater than the corresponding adjustment threshold, the corresponding module of the given pipeline defect detection device is not optimized and adjusted.

[0059] S520, when any one of the real-time flow rate and the real-time jitter data is greater than the corresponding adjustment threshold, the corresponding module of the target item greater than the corresponding adjustment threshold is adjusted to the corresponding adjustment threshold, until it is less than the corresponding adjustment threshold and the difference between the corresponding adjustment threshold is in a preset anti-oscillation threshold interval. For example, the corresponding module of the flow rate is the resistance adjustment module, and the corresponding module of the jitter is the shock absorber installed in the device.

[0060] The above, only when the real-time flow rate and the real-time jitter data are greater than the corresponding adjustment threshold, the adjustment is performed, which reduces the adjustment cost and ensures that the detection device has a higher detection accuracy, and a anti-oscillation threshold interval is set, a smoother and more robust adjustment strategy is introduced, the adjustment action is prevented from being performed back and forth, and the stability of the control system is improved.

[0061] Further, the method further comprises detection of a water leakage point, including the following steps:

[0062] S10, detecting the frequency of the water leakage sound wave in real time by using a Doppler effect based frequency distortion model in the given pipeline defect detection device.

[0063] S20, determining the water leakage point position according to the measured frequency inversion point based on the preset pipeline map, and it can be understood that the detected sound wave frequency gradually decreases when the detection device drives to the water leakage point, and the detected sound wave frequency gradually increases when the detection device drives away from the water leakage point.

[0064] S30, constructing a sound wave frequency curve according to the frequency of the water leakage sound wave detected in real time when the shape of the given pipeline is a non-regular straight line, wherein the sound wave frequency curve represents the relationship between the sound wave frequency and the position of the given pipeline defect detection device, and the position of the given pipeline defect detection device is obtained by the pipeline map and the corresponding positioning information of the given pipeline defect detection device in the pipeline map.

[0065] S40, determining the water leakage point position based on the maximum trend inversion point in the sound wave frequency curve, and in a specific embodiment, the position information of each inversion point can be determined by combining the pipeline map and the driving position of the detection device in the pipeline map to verify the water leakage point position.

[0066] In the above, considering that the given pipeline is a non-regular straight line, for example, an S-shaped or zigzag-shaped line, the frequency curve may also be inverted at the turning point, and the present application does not focus on the details of the curve, but on the macro trend inversion point, so that the determined water leakage point position is more accurate.

[0067] The embodiment of the present application also provides a non-transitory computer readable storage medium, which can be arranged in an electronic device to save at least one instruction or at least one program related to a method in the method embodiment, and the at least one instruction or the at least one program is loaded and executed by the processor to realize the method provided by the above-mentioned embodiment.

[0068] The embodiment of the present application also provides an electronic device, which comprises a processor and the aforementioned non-transitory computer readable storage medium.

[0069] Although some specific embodiments of the present application have been described in detail by examples, those skilled in the art should understand that the above examples are only for illustration, but not for limiting the scope of the present application. Those skilled in the art should also understand that various modifications can be made to the embodiments without departing from the scope and spirit of the present application. The scope of the present application is defined by the appended claims.

Claims

1. An optimized method for pipeline defect detection, characterized in that, The method includes the following steps: Based on the detection videos captured by the given pipeline defect detection device at each preset flow rate corresponding to the water flow in the preset pipeline, the detection information of several preset defect types corresponding to each preset flow rate is obtained. Based on the detection information of several preset defect types and the actual defect information corresponding to a given pipeline, a target fitting curve between the first defect detection score and the flow velocity is constructed. The target points with defect detection scores not lower than the first preset score threshold are determined from the target fitting curve, and the flow rate is extended from the target points to determine the flow rate range. Based on n sets of preset jitter data, n sets of defect detection information corresponding to each preset flow rate within the flow rate range are collected, and the second defect detection score corresponding to each set of defect detection information is obtained, thus obtaining the correspondence between each set of jitter data and the second defect detection score under each preset flow rate. Specifically, the second defect detection score corresponding to each set of defect detection information is obtained through the following steps: For any preset flow rate within the flow rate range, based on the n sets of defect detection information corresponding to the preset flow rate, combined with the actual defect information corresponding to the given pipeline, the second identification proportion of the preset defect type under each set of defect detection information, and the second identification score corresponding to each identified preset defect type are calculated; the second identification proportion and several second identification scores corresponding to each set of defect detection information are weighted and summed to obtain the second defect detection score corresponding to each set of defect detection information. Based on the aforementioned correspondence, target flow rate and target jitter data are determined from a number of preset flow rates and preset jitter data whose corresponding second defect detection scores are not lower than the second preset score threshold. These target flow rates and target jitter data are then used as adjustment thresholds to optimize the corresponding modules of a given pipeline defect detection device. The target flow rate and target jitter data are determined through the following steps: An m×n detection score matrix is ​​constructed based on the second defect detection scores corresponding to n sets of preset jitter data at each preset flow rate; where m is the number of preset flow rates within the flow rate range. Preset flow rate and preset jitter data association pairs corresponding to a number of second defect detection scores not lower than the second preset score threshold are determined from the detection score matrix. Several association pairs are traversed, and the number of preset flow rates corresponding to each preset jitter data is obtained from the association pairs. The largest jitter data among a number of preset jitter data whose proportion of the number of corresponding preset flow rates is not less than the proportion of the number of preset flow rates is determined as the target jitter data. Several association pairs are traversed, and the largest preset flow rate corresponding to the target jitter data is obtained from the association pairs and determined as the target flow rate.

2. The pipeline defect detection optimization method according to claim 1, characterized in that, The target fitting curve is constructed using the following steps: Based on the detection information of several preset defect types and the actual defect information corresponding to a given pipeline, calculate the first identification percentage of the preset defect type under each preset flow rate, and the first identification score corresponding to each preset defect type identified. The weighted sum of the first identification percentage and several first identification scores corresponding to any preset flow rate is calculated to obtain the first defect detection score corresponding to any preset flow rate. Based on the first defect detection score corresponding to each preset flow rate, a target fitting curve between the first defect detection score and the flow rate is constructed.

3. The pipeline defect detection optimization method according to claim 2, characterized in that, The first identification score corresponding to each preset defect type is obtained through the following steps: For any given defect type, based on several preset feature labels, the detection information of quantifiable features and non-quantifiable features of the defect type are determined from the detection information of the defect type. For quantifiable features, the detection information of the quantifiable features is calculated and the actual defect information of the corresponding preset defect type of the given pipeline is calculated, and the defect similarity corresponding to the quantifiable features is calculated. For non-quantitative features, the semantic similarity corresponding to the non-quantitative features is calculated based on the detection information of the non-quantitative features and the actual defect information of the given pipeline corresponding to the preset defect type. The defect similarity corresponding to quantifiable features and the semantic similarity corresponding to non-quantifiable features are input into a pre-trained machine learning model to obtain the first recognition score for each preset defect type.

4. The pipeline defect detection optimization method according to claim 1, characterized in that, The process of extending the flow velocity at the target point to determine the flow velocity range includes the following steps: When the number of preset flow velocities that are not greater than the flow velocity corresponding to the target point is greater than the preset number threshold, the target point is extended in the direction of decreasing flow velocity, and the flow velocity range is determined based on the maximum and minimum values ​​of several preset flow velocities corresponding to the flow velocity from 0 to the new target point. When the number of preset flow velocities that are not greater than the flow velocity corresponding to the target point is not greater than the preset number threshold, the flow velocity range is determined based on the maximum and minimum values ​​among several preset flow velocities that are not greater than the flow velocity corresponding to the target point.

5. The pipeline defect detection optimization method according to claim 1, characterized in that, The optimization adjustment of the corresponding module of a given pipeline defect detection device by using target flow velocity and target jitter data as adjustment thresholds includes the following steps: When the real-time flow rate and real-time jitter data detected by the given pipeline defect detection device are not greater than the corresponding adjustment threshold, the corresponding module of the given pipeline defect detection device will not be optimized or adjusted. When either the real-time flow rate or the real-time jitter data exceeds the corresponding adjustment threshold, the corresponding module of the target item that exceeds the corresponding adjustment threshold is adjusted to the corresponding adjustment threshold until it is less than the corresponding adjustment threshold and the difference between the two adjustment thresholds is within the preset anti-oscillation threshold range.

6. The pipeline defect detection optimization method according to claim 1, characterized in that, The method further includes the following steps: The frequency of the sound wave of water leakage is detected in real time by using a frequency distortion model based on the Doppler effect in a given pipeline defect detection device. Based on a preset pipeline map, if the given pipeline has a regular straight line shape, the location of the leak can be determined according to the measured frequency reversal point. If the given pipeline has an irregular straight shape, a sound wave frequency curve is constructed based on the frequency of the leaking sound waves detected in real time. The sound wave frequency curve represents the relationship between the sound wave frequency and the location of the given pipeline defect detection device. The location of the given pipeline defect detection device is obtained through the pipeline map and the corresponding positioning information of the given pipeline defect detection device in the pipeline map. The location of the leak is determined based on the point of maximum trend reversal in the sound wave frequency curve.

7. A non-transitory computer-readable storage medium, wherein the storage medium stores at least one instruction or at least one program segment, characterized in that, The at least one instruction or the at least one program segment is loaded and executed by the processor to implement the pipeline defect detection optimization method as described in any one of claims 1-6.

8. An electronic device, characterized in that, Includes a processor and the non-transitory computer-readable storage medium as described in claim 7.

Images