CN118154588A - Large-diameter pressure steel pipe quality detection method and system based on contour extraction - Google Patents
Large-diameter pressure steel pipe quality detection method and system based on contour extraction Download PDFInfo
- Publication number
- CN118154588A CN118154588A CN202410565741.8A CN202410565741A CN118154588A CN 118154588 A CN118154588 A CN 118154588A CN 202410565741 A CN202410565741 A CN 202410565741A CN 118154588 A CN118154588 A CN 118154588A
- Authority
- CN
- China
- Prior art keywords
- cloud data
- point cloud
- abnormal point
- abnormal
- acquiring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 229910000831 Steel Inorganic materials 0.000 title claims abstract description 48
- 239000010959 steel Substances 0.000 title claims abstract description 48
- 238000001514 detection method Methods 0.000 title claims abstract description 45
- 238000000605 extraction Methods 0.000 title claims abstract description 25
- 230000002159 abnormal effect Effects 0.000 claims abstract description 268
- 230000002776 aggregation Effects 0.000 claims abstract description 88
- 238000004220 aggregation Methods 0.000 claims abstract description 88
- 230000007547 defect Effects 0.000 claims abstract description 84
- 238000000034 method Methods 0.000 claims abstract description 58
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 20
- 230000006870 function Effects 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 238000013135 deep learning Methods 0.000 claims description 6
- 238000003062 neural network model Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 4
- 238000002604 ultrasonography Methods 0.000 claims description 4
- 238000007689 inspection Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000010191 image analysis Methods 0.000 abstract description 2
- 239000000523 sample Substances 0.000 description 14
- 230000008569 process Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 239000000758 substrate Substances 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
Landscapes
- Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)
Abstract
The invention relates to the technical field of image analysis, in particular to a large-diameter penstock quality detection method and system based on contour extraction. The method comprises the steps of obtaining abnormal point cloud data of a large-diameter pressure steel pipe; acquiring an aggregation degree value according to the distribution condition of abnormal point cloud data of the same position under the detection position and the distance between each abnormal point cloud data and adjacent abnormal point cloud data; acquiring an aggregation consistent value according to the quantity of the reference abnormal point cloud data of the abnormal point cloud data and the fluctuation condition of the aggregation degree value, further acquiring a neighborhood distance threshold value in a DBSCAN clustering algorithm, and clustering the abnormal point cloud data into a final cluster; and determining defects according to the aggregation consistent value and the coordinate point of the abnormal point cloud data in the final cluster, and detecting the quality of the large-diameter penstock. According to the invention, the defects in the large-diameter pressure steel pipe are determined by analyzing the distribution of the abnormal point cloud data, so that the quality of the large-diameter pressure steel pipe is accurately detected.
Description
Technical Field
The invention relates to the technical field of image analysis, in particular to a large-diameter penstock quality detection method and system based on contour extraction.
Background
The large-diameter pressure steel pipe is a pipeline with special structure and function and is used for conveying a large amount of liquid or gas, a certain negative pressure is required to be provided, and a suction effect is achieved so that the liquid or gas can smoothly flow, so that the quality of the large-diameter pressure steel pipe is required to be qualified, and the normal operation of the large-diameter pressure steel pipe is ensured.
In the existing method, the saw-tooth-shaped continuous ultrasonic flaw detection is carried out on the large-diameter pressure steel pipe through an ultrasonic flaw detection device, a three-dimensional point cloud data model of the large-diameter pressure steel pipe is built, a defect area in the large-diameter pressure steel pipe is determined, and the quality of the large-diameter pressure steel pipe is detected. However, in actual situations, the point cloud data acquired by the ultrasonic flaw detection device can generate corresponding changes along with the serrated movement of the probe, errors are likely to occur in the point cloud data in the process of acquiring the point cloud data by the movement of the probe, so that a defect area in the large-diameter penstock cannot be accurately identified according to the point cloud data, further, the quality of the large-diameter penstock cannot be accurately detected, and stable working of the large-diameter penstock cannot be ensured.
Disclosure of Invention
In order to solve the technical problem that the point cloud data are easy to generate errors in the process of acquiring the point cloud data by moving a probe, so that a defect area in a large-diameter pressure steel pipe cannot be accurately identified according to the point cloud data, and further the quality of the large-diameter pressure steel pipe cannot be accurately detected, the invention aims to provide a large-diameter pressure steel pipe quality detection method and system based on contour extraction, and the adopted technical scheme is as follows:
In a first aspect, an embodiment of the present invention provides a method for detecting quality of a large-diameter penstock based on contour extraction, the method comprising the steps of:
acquiring abnormal point cloud data of the large-diameter pressure steel pipe at each detection position;
classifying the outlier point cloud data into point cloud data categories based on distances between outlier point cloud data; optionally selecting one piece of abnormal point cloud data as target point cloud data, and constructing the abnormal point cloud data of the same position as the target point cloud data in all detection positions as a point cloud data set; acquiring an aggregation degree value of the target point cloud data according to the distribution condition of the abnormal point cloud data in the point cloud data set in the point cloud data category to which the target point cloud data belong and the distance between the target point cloud data and the preset number of adjacent abnormal point cloud data;
Screening out reference abnormal point cloud data of the target point cloud data according to the distance between the target point cloud data and each other abnormal point cloud data in the class of the target point cloud data to which the target point cloud data belongs; acquiring an aggregation consistent value of the target point cloud data according to the quantity of the reference abnormal point cloud data and the fluctuation condition of the aggregation degree value;
Acquiring a neighborhood distance threshold value in a DBSCAN clustering algorithm according to the aggregation consistent value of each abnormal point cloud data, and clustering the abnormal point cloud data to acquire a final cluster; acquiring coordinate points of final defect point cloud data according to the aggregation consistent value and the coordinate points of each abnormal point cloud data in each final cluster, further determining defects in the large-diameter pressure steel pipe, and detecting the quality of the large-diameter pressure steel pipe;
the method for obtaining the defects in the large-diameter pressure steel pipe comprises the following steps: and inputting coordinate points of all final defect point cloud data into a trained deep learning neural network model, and accurately outputting defects in the large-diameter penstock.
Further, the method for obtaining the aggregation level value comprises the following steps:
acquiring the number of abnormal point cloud data in the point cloud data set in the point cloud data category to which the target point cloud data belong as a first number of the target point cloud data;
And acquiring the aggregation degree value of the target point cloud data according to the first quantity, the ratio of the total quantity of all abnormal point cloud data in the point cloud data set to the first quantity and the Euclidean distance between the target point cloud data and the preset quantity of adjacent abnormal point cloud data.
Further, the calculation formula of the aggregation level value is as follows:
; in the/> The aggregation degree value of the kth abnormal point cloud data is obtained; The total number of all abnormal point cloud data in the point cloud data set where the kth abnormal point cloud data is located; /(I) A first quantity of kth outlier cloud data; m is a preset number; /(I)The Euclidean distance between the kth abnormal point cloud data and the coordinate point corresponding to the mth adjacent abnormal point cloud data; exp is an exponential function based on a natural constant.
Further, the method for acquiring the reference abnormal point cloud data comprises the following steps:
acquiring Euclidean distances between the target point cloud data and coordinate points corresponding to other abnormal point cloud data in the class of the target point cloud data, wherein the target point cloud data belong to the target point cloud data, and the Euclidean distances are used as first distances;
and taking other abnormal point cloud data of the non-target point cloud data corresponding to the minimum first distance as reference abnormal point cloud data of the target point cloud data.
Further, the calculation formula of the aggregate consistent value is:
; in the/> An aggregation consistent value for the kth outlier cloud data; /(I)The total number of reference outlier cloud data that is kth outlier cloud data; /(I)The aggregation degree value of the first reference abnormal point cloud data of the kth abnormal point cloud data; /(I)The average value of the aggregation degree values of all the reference abnormal point cloud data of the kth abnormal point cloud data; /(I)As a function of absolute value; norm is a normalization function; /(I)Is a first preset constant, greater than 0.
Further, the method for obtaining the neighborhood distance threshold value in the DBSCAN clustering algorithm according to the aggregation consistent value of each abnormal point cloud data comprises the following steps:
Taking the abnormal point cloud data corresponding to the maximum aggregation consistent value as characteristic point cloud data;
And acquiring the average value of the first distances between all the characteristic point cloud data and the reference abnormal point cloud data thereof as a neighborhood distance threshold value in the DBSCAN clustering algorithm.
Further, the method for obtaining the coordinate point of the final defect point cloud data according to the aggregate consistent value and the coordinate point of each abnormal point cloud data in each final cluster comprises the following steps:
Acquiring a matching degree value of each abnormal point cloud data according to the difference of the aggregation consistent value of each abnormal point cloud data and the average value of the aggregation consistent values of all the abnormal point cloud data in the final cluster to which the abnormal point cloud data belongs; wherein the difference and the matching degree value are in a negative correlation relationship;
multiplying the matching degree value of each abnormal point cloud data with the corresponding coordinate point to obtain a corrected coordinate point corresponding to each abnormal point cloud data;
And acquiring the central coordinate points of the corrected coordinate points corresponding to all the abnormal point cloud data in each final cluster, and taking the central coordinate points as the coordinate points of each final defect point cloud data.
Further, the method for acquiring the point cloud data category comprises the following steps:
And acquiring Euclidean distance of coordinate points corresponding to any two abnormal point cloud data, and clustering the abnormal point cloud data by using an OPTICS density clustering algorithm to obtain the point cloud data category.
Further, the method for acquiring the abnormal point cloud data comprises the following steps:
Acquiring point cloud data of the large-diameter pressure steel pipe at each detection position; wherein each point cloud data comprises ultrasound data and a reference distance;
and acquiring an ultrasonic data threshold value under each reference distance, and when the ultrasonic data of the point cloud data is larger than the ultrasonic data threshold value under the corresponding reference distance, the corresponding point cloud data is abnormal point cloud data.
In a second aspect, another embodiment of the present invention provides a large diameter penstock quality inspection system based on profile extraction, the system comprising: a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of any one of the methods described above when executing the computer program.
The invention has the following beneficial effects:
Dividing the abnormal point cloud data into point cloud data categories based on the distance between the abnormal point cloud data, and primarily dividing the abnormal point cloud data, so that the subsequent analysis of the aggregation degree of each abnormal point cloud data is facilitated; for the sake of clarity, further selecting an abnormal point cloud data as the target point cloud data, constructing the abnormal point cloud data at the same position as the target point cloud data under all detection positions as a point cloud data set, and making accuracy for accurately analyzing the aggregation degree of the target point cloud data; obtaining the aggregation degree value of the target point cloud data according to the distribution condition of the abnormal point cloud data in the point cloud data set in the point cloud data category to which the target point cloud data belong and the distance between the target point cloud data and the preset number of adjacent abnormal point cloud data, and primarily judging the possibility that the target point cloud data are true defect point cloud data; in order to further analyze the possibility degree of each abnormal point cloud data as the true defect point cloud data, further screening out reference abnormal point cloud data of the target point cloud data according to the distance between the target point cloud data and each other abnormal point cloud data in the class of the target point cloud data, and improving the accuracy and efficiency of analyzing the target point cloud data as the true defect point cloud data; further, according to the quantity of the reference abnormal point cloud data and the fluctuation condition of the aggregation degree value, an aggregation consistent value of the target point cloud data is obtained, and the possibility that the target point cloud data is true defect point cloud data is further determined; in order to accurately acquire defects in the large-diameter penstock, further, according to the aggregation consistent value of each abnormal point cloud data, accurately acquiring a neighborhood distance threshold value in a DBSCAN clustering algorithm, further, clustering the abnormal point cloud data to acquire a final cluster, accurately determining coordinate points of the final defect point cloud data, further, according to the aggregation consistent value and the coordinate points of each abnormal point cloud data in each final cluster, accurately acquiring the coordinate points of the final defect point cloud data, and reducing influence caused by deviation of the abnormal point cloud data; and then accurately determining defects in the large-diameter pressure steel pipe based on coordinate points of final defect point cloud data, accurately detecting the quality of the large-diameter pressure steel pipe, and ensuring stable working of the large-diameter pressure steel pipe.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic flow chart of a method for detecting quality of a large-diameter penstock based on contour extraction according to an embodiment of the invention.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following is a detailed description of specific implementation, structure, characteristics and effects of the large-diameter penstock quality detection method and system based on contour extraction according to the invention with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The invention provides a specific scheme of a large-diameter penstock quality detection method based on contour extraction, which is specifically described below with reference to the accompanying drawings.
The aim of the embodiment of the invention is as follows: when the point cloud data of the large-diameter pressure steel pipe at each detection position are obtained through the ultrasonic flaw detection device, errors may exist in the obtained point cloud data, and therefore defective areas in the large-diameter pressure steel pipe are inaccurate in identification, and quality of the large-diameter pressure steel pipe cannot be accurately detected.
Referring to fig. 1, a flow chart of a method for detecting quality of a large-diameter penstock based on contour extraction according to an embodiment of the invention is shown, the method comprises the following steps:
Step S1: and acquiring abnormal point cloud data of the large-diameter penstock at each detection position.
Specifically, an optional section of large-diameter penstock is taken as an example in the embodiment of the invention, wherein the length of the selected large-diameter penstock is necessarily greater than 3 meters, a plurality of detection positions are set on the surface of the large-diameter penstock according to a preset saw-tooth scanning path and angle, the path distance between adjacent detection positions is set to be 3 meters in the embodiment of the invention, and an operator can set the path distance between the adjacent detection positions according to actual conditions, so that the method is not limited. The method comprises the steps of acquiring point cloud data of a large-diameter pressure steel pipe at each detection position through an ultrasonic flaw detection device, wherein the number of the point cloud data at each detection position is the same, and when the probe of the ultrasonic flaw detection device acquires the point cloud data of the large-diameter pressure steel pipe at each detection position, the scanning paths of the probe on the large-diameter pressure steel pipe are the same, namely the point cloud data at the same position at each detection position corresponds to the same area of the large-diameter pressure steel pipe. Each point cloud data comprises ultrasonic data and a reference distance, wherein the ultrasonic data represents an ultrasonic reflection value of a point corresponding to the point cloud data, and the reference distance represents a distance between the point corresponding to the point cloud data and a detection position. When the surface of the large-diameter penstock has no defect, the ultrasonic data can change along with the change of the reference distances, namely, each reference distance corresponds to an ultrasonic data threshold value, and when the ultrasonic data of the point cloud data is larger than the ultrasonic data threshold value under the corresponding reference distance, the region in the large-diameter penstock corresponding to the corresponding point cloud data is possibly a defect region, so that the corresponding point cloud data is taken as abnormal point cloud data. To this end, outlier cloud data at each detection position is determined. The method of obtaining the ultrasound data threshold value at each reference distance may refer to the method in https:// zhuanlan. Zhihu. Com/p/577170349 utm_id=0, and it should be noted that the content in https:// zhuanlan. Zhihu. Com/p/577170349 utm_id=0 is a known technique, so the method of obtaining the ultrasound data threshold value at each reference distance is a known technique and will not be described in detail.
Step S2: classifying the outlier point cloud data into point cloud data categories based on distances between outlier point cloud data; optionally selecting one piece of abnormal point cloud data as target point cloud data, and constructing the abnormal point cloud data of the same position as the target point cloud data in all detection positions as a point cloud data set; and acquiring the aggregation degree value of the target point cloud data according to the distribution condition of the abnormal point cloud data in the point cloud data set in the point cloud data category to which the target point cloud data belong and the distance between the target point cloud data and the preset number of adjacent abnormal point cloud data.
Specifically, when the probe moves at the detection position, the point cloud data corresponding to the normal area of the surface of the large-diameter penstock has the condition that the abnormal point cloud data is mistaken, meanwhile, the point cloud data corresponding to the defect area of the surface of the large-diameter penstock may deviate in the process of acquisition, but the overall distribution condition of the abnormal point cloud data corresponding to a certain defect point on the surface of the large-diameter penstock is certainly intensively distributed, so that when coordinate points corresponding to the abnormal point cloud data at the same position at the detection position are gathered, the area of the large-diameter penstock corresponding to the abnormal point cloud data at the same position at the detection position is more likely to be the defect area. Therefore, in the embodiment of the invention, all abnormal point cloud data are mapped into a three-dimensional sample space, wherein the three-dimensional sample space is scaled in an equal proportion to the actual detection position of the large-diameter penstock, and the coordinate origin and the coordinate axis direction in the three-dimensional sample space can be set according to the actual situation without limitation. The Euclidean distance of coordinate points corresponding to any two abnormal point cloud data in the three-dimensional sample space is obtained, the abnormal point cloud data are clustered through an OPTICS density clustering algorithm, a point cloud data class is obtained, each abnormal point cloud data in each point cloud data class is analyzed, the aggregation degree value of each abnormal point cloud data is obtained, and the possibility that each abnormal point cloud data is true defect point cloud data is primarily judged. The OPTICS density clustering algorithm and the Euclidean distance obtaining method are known techniques, and will not be described in detail. In order to clearly illustrate a method for acquiring the aggregation level value of each abnormal point cloud data, in the embodiment of the present invention, one abnormal point cloud data is optionally selected as the target point cloud data, and the target point cloud data is taken as an example for illustration. Firstly, constructing abnormal point cloud data of the same position as the target point cloud data in all detection positions as a point cloud data set, for example, if the target point cloud data is the ith point cloud data in the detection positions, constructing the point cloud data of which the ith point cloud data is the abnormal point cloud data in all detection positions as a point cloud data set. When the abnormal point cloud data in the point cloud data set is more, the point cloud data corresponding to the defect point of the large-diameter penstock is more likely to be indicated as the target point cloud data. Secondly, taking the point cloud data category to which the target point cloud data belongs as the target point cloud data category, and when the distribution of abnormal point cloud data in the point cloud data set to which the target point cloud data belongs in the target point cloud data category is more, indicating that the aggregation degree of the target point cloud data is larger, indirectly indicating that the target point cloud data is more likely to be the point cloud data corresponding to the defect point of the large-diameter penstock. Further acquiring Euclidean distances between coordinate points corresponding to the target point cloud data and coordinate points corresponding to the preset number of adjacent abnormal point cloud data around the coordinate points in the three-dimensional sample space, and when the Euclidean distances between the target point cloud data and the preset number of adjacent abnormal point cloud data are smaller, indicating that the distribution of the target point cloud data is more concentrated, so that the concentration degree value of the target point cloud data is acquired according to the distribution condition of the abnormal point cloud data in the point cloud data set of the target point cloud data in the point cloud data category to which the target point cloud data belongs and the distances between the target point cloud data and the preset number of adjacent abnormal point cloud data. The larger the aggregation degree value is, the more likely the target point cloud data is the point cloud data corresponding to the defect point of the large-diameter penstock.
As an example, taking kth abnormal point cloud data as an example, the kth abnormal point cloud data is target point cloud data. The point cloud data category of the kth abnormal point cloud data is the target point cloud data category, the number of the abnormal point cloud data in the point cloud data set of the kth abnormal point cloud data in the target point cloud data category is obtained and is used as the first number of the kth abnormal point cloud data, when the first number is larger, the distribution of the kth abnormal point cloud data is more concentrated, and the kth abnormal point cloud data is more likely to be the point cloud data corresponding to the defect point of the large-diameter penstock. When the total quantity of all abnormal point cloud data in the point cloud data set where the first quantity and the kth abnormal point cloud data are located is more similar, further describing the more concentrated distribution of the kth abnormal point cloud data. According to the embodiment of the invention, the preset number is set to be 5, an implementer can set the preset number according to the actual situation, the implementation is not limited in this case, the Euclidean distance between the kth abnormal point cloud data and the nearest 5 abnormal point cloud data in the three-dimensional sample space is obtained, wherein the nearest 5 abnormal point cloud data of the kth abnormal point cloud data in the three-dimensional sample space is the 5 abnormal point cloud data with the coordinate point nearest to the coordinate point of the kth abnormal point cloud data, and when the Euclidean distance between the kth abnormal point cloud data and the nearest 5 abnormal point cloud data in the three-dimensional sample space is smaller, the aggregation degree of the kth abnormal point cloud data is larger, and the point cloud data corresponding to the defect point of the large-diameter pressure steel pipe is indirectly indicated. Therefore, according to the first number of the abnormal point cloud data, the ratio of the total number of all abnormal point cloud data in the point cloud data set where the kth abnormal point cloud data is located to the first number of the abnormal point cloud data, and the euclidean distance between the kth abnormal point cloud data and the preset number of adjacent abnormal point cloud data, a calculation formula for obtaining the aggregation degree value of the kth abnormal point cloud data is as follows:
; in the/> The aggregation degree value of the kth abnormal point cloud data is obtained; The total number of all abnormal point cloud data in the point cloud data set where the kth abnormal point cloud data is located; /(I) A first quantity of kth outlier cloud data; m is a preset number, and the embodiment of the invention is set to be 5; /(I)The Euclidean distance between the kth abnormal point cloud data and the coordinate point corresponding to the mth adjacent abnormal point cloud data; exp is an exponential function based on a natural constant.
It should be noted that the number of the substrates,The larger the abnormal point cloud data in the point cloud data set of the kth abnormal point cloud data is, the more the abnormal point cloud data in the point cloud data set of the kth abnormal point cloud data is gathered, the greater the gathering degree of the kth abnormal point cloud data is, and the greater the gathering degree of the kth abnormal point cloud data isThe larger; /(I)Smaller,/>The closer/>Wherein/>The more the abnormal point cloud data in the point cloud data set where the kth abnormal point cloud data is located in the target point cloud data category is concentrated, the more the kth abnormal point cloud data is likely to be the point cloud data corresponding to the defect point of the large-diameter pressure steel pipe is indirectly indicated,The larger the/>The larger; /(I)The smaller the data, the greater the aggregation degree of the kth abnormal point cloud data is, namely/>The larger the/>The larger; thus,/>The larger the k-th abnormal point cloud data is, the more likely the k-th abnormal point cloud data is the point cloud data corresponding to the defect point of the large-diameter penstock.
And acquiring the aggregation degree value of each abnormal point cloud data according to the method for acquiring the aggregation degree value of the kth abnormal point cloud data.
Step S3: screening out reference abnormal point cloud data of the target point cloud data according to the distance between the target point cloud data and each other abnormal point cloud data in the class of the target point cloud data to which the target point cloud data belongs; and acquiring an aggregation consistent value of the target point cloud data according to the quantity of the reference abnormal point cloud data and the fluctuation condition of the aggregation degree value.
Specifically, in order to reduce the influence caused by the deviation of the point cloud data, the embodiment of the invention acquires the Euclidean distance between the target point cloud data and the coordinate points corresponding to other abnormal point cloud data in the class of the point cloud data to which the target point cloud data belongs, and the Euclidean distance is used as the first distance, when the first distance is smaller, the distribution of the target point cloud data is more concentrated, and the target point cloud data is more likely to be the point cloud data corresponding to the defect point of the large-diameter penstock. In order to accurately determine the possibility that the target point cloud data is the true defect point cloud data, other abnormal point cloud data of non-target point cloud data corresponding to the minimum first distance are taken as reference abnormal point cloud data of the target point cloud data, the target point cloud data and the reference abnormal point cloud data of the target point cloud data are point cloud data corresponding to the same defect point of the large-diameter pressure steel pipe, when the number of the reference abnormal point cloud data of the target point cloud data is larger, the target point cloud data and the reference point cloud data are more aggregated, and the target point cloud data are more likely to be the true defect point cloud data indirectly; when the aggregation degree values of the reference abnormal point cloud data of the target point cloud data are more similar, further explaining that the target point cloud data are more likely to be true defect point cloud data. Therefore, according to the quantity of the reference abnormal point cloud data of the target point cloud data and the fluctuation condition of the aggregation degree value, the aggregation consistent value of the target point cloud data is obtained, wherein the larger the aggregation consistent value is, the more likely the target point cloud data is the true defect point cloud data.
As an example, taking the kth abnormal point cloud data in step S2 as an example, the kth abnormal point cloud data is the target point cloud data. The calculation formula for acquiring the aggregation consistent value of the kth abnormal point cloud data is as follows:
; in the/> An aggregation consistent value for the kth outlier cloud data; /(I)The total number of reference outlier cloud data that is kth outlier cloud data; /(I)The aggregation degree value of the first reference abnormal point cloud data of the kth abnormal point cloud data; /(I)The average value of the aggregation degree values of all the reference abnormal point cloud data of the kth abnormal point cloud data; /(I)As a function of absolute value; norm is a normalization function; /(I)Is a first preset constant, greater than 0.
Embodiments of the invention willSet to 1, avoid denominator to 0, and the practitioner can set/>, according to the actual situationIs not limited herein.
It should be noted that the number of the substrates,The larger the distribution of kth abnormal point cloud data is, the more likely to be concentrated, the more likely the kth abnormal point cloud data is to be true defect point cloud data,/>The larger; /(I)Smaller,/>And/>The more similar,/>The smaller the reference abnormal point cloud data of the kth abnormal point cloud data is, the more likely the reference abnormal point cloud data is the point cloud data corresponding to the same defect point of the large-diameter pressure steel pipe, and the more likely the kth abnormal point cloud data is the true defect point cloud data is indirectly indicated.
And acquiring the aggregation consistent value of each abnormal point cloud data according to the method for acquiring the aggregation consistent value of the kth abnormal point cloud data.
Step S4: acquiring a neighborhood distance threshold value in a DBSCAN clustering algorithm according to the aggregation consistent value of each abnormal point cloud data, and clustering the abnormal point cloud data to acquire a final cluster; and acquiring coordinate points of the final defect point cloud data according to the aggregation consistent value and the coordinate points of each abnormal point cloud data in each final cluster, further determining defects in the large-diameter pressure steel pipe, and detecting the quality of the large-diameter pressure steel pipe.
Specifically, the larger the aggregation consistent value is, the more likely the corresponding abnormal point cloud data is the true defect point cloud data, so that the abnormal point cloud data corresponding to the maximum aggregation consistent value is used as the characteristic point cloud data in the embodiment of the invention. The maximum aggregation consistent value at least corresponds to one abnormal point cloud data. In order to improve the accuracy of gathering abnormal point cloud data of the same defect point of the large-diameter penstock in the same area, further accurately acquiring a coordinate point of final defect point cloud data, and determining defects in the large-diameter penstock. The more likely each characteristic point cloud data is that of the same defect point of the large-diameter penstock with respect to the abnormal point cloud data, so that in order to gather the abnormal point cloud data of the same defect point of the large-diameter penstock to the greatest extent, the embodiment of the invention obtains the first distance between each characteristic point cloud data and the abnormal point cloud data with respect to the characteristic point cloud data as the characteristic distance of the corresponding characteristic point cloud data, wherein the characteristic distances of different characteristic point cloud data are different, and in order to accurately analyze each abnormal point cloud data, the embodiment of the invention obtains the average value of all the characteristic distances as the neighborhood distance threshold value in the DBSCAN clustering algorithm, the clustering accuracy of the DBSCAN clustering algorithm is improved, and further, abnormal point cloud data are clustered through the DBSCAN clustering algorithm according to the Euclidean distance between any two abnormal point cloud data corresponding to coordinate points, so that a final clustering cluster is obtained. In the embodiment of the invention, the minimum number in the DBSCAN clustering algorithm is set to be 5, and an implementer can set the minimum number in the DBSCAN clustering algorithm according to actual conditions, and the method is not limited. The DBSCAN clustering algorithm is a well-known technique, and will not be described in detail. At this time, the abnormal point cloud data in each final cluster is defaulted to point cloud data corresponding to the same defect point in the large-diameter penstock. In order to accurately determine defects in large-diameter penstock, the embodiment of the invention acquires the average value of the aggregation consistent values of all abnormal point cloud data in each final cluster, and uses the average value as the aggregation standard value of each final cluster, when the aggregation consistent value of certain abnormal point cloud data is more similar to the aggregation standard value of the final cluster to which the abnormal point cloud data belongs, the abnormal point cloud data is more likely to be true defect point cloud data, and the abnormal point cloud data corresponds to a coordinate point more accurately. Therefore, when the difference between the aggregation consistent value of a certain abnormal point cloud data and the average value of the aggregation consistent values of all abnormal point cloud data in the final cluster to which the certain abnormal point cloud data belongs is smaller, the abnormal point cloud data is more likely to be real defect point cloud data, and the matching degree value of the abnormal point cloud data is larger. In order to accurately acquire defects in the large-diameter penstock, the embodiment of the invention corrects coordinate points corresponding to each abnormal point cloud data through the matching degree value of each abnormal point cloud data to acquire corrected coordinates corresponding to each abnormal point cloud data, and further accurately acquires coordinate points of each final defect point cloud data according to corrected coordinate points corresponding to all abnormal point cloud data in each final cluster, accurately determines defects in the large-diameter penstock, and accurately detects the quality of the large-diameter penstock. The specific method for detecting the mass of the large-diameter pressure steel pipe comprises the following steps of:
(1) And obtaining a matching degree value.
Preferably, the method for obtaining the matching degree value is as follows: acquiring a matching degree value of each abnormal point cloud data according to the difference of the aggregation consistent value of each abnormal point cloud data and the average value of the aggregation consistent values of all the abnormal point cloud data in the final cluster to which the abnormal point cloud data belongs; wherein the difference and the matching degree value are in a negative correlation.
Taking the kth abnormal point cloud data in the step S2 as an example, a calculation formula for obtaining the matching degree value of the kth abnormal point cloud data is as follows:
; in the/> The matching degree value of the kth abnormal point cloud data is the matching degree value of the kth abnormal point cloud data; /(I)An aggregation consistent value for the kth outlier cloud data; /(I)The average value of the aggregation consistent values of all the abnormal point cloud data in the final cluster to which the kth abnormal point cloud data belongs; /(I)As a function of absolute value; exp is an exponential function with natural constant as a base; /(I)And the difference is the difference of the aggregation consistent value of the kth abnormal point cloud data and the average value of the aggregation consistent values of all abnormal point cloud data in the final cluster to which the kth abnormal point cloud data belongs.
It should be noted that the number of the substrates,The smaller, i.e./>The more prone to 1,/>And/>The more similar the kth abnormal point cloud data is, the more likely the kth abnormal point cloud data is the true defect point cloud data, the more accurate the coordinate point corresponding to the kth abnormal point cloud data is,/>The larger; thus,/>The larger the k-th abnormal point cloud data is, the more likely the k-th abnormal point cloud data is to be the true defect point cloud data, and the more accurate the k-th abnormal point cloud data corresponds to the coordinate point. In other embodiments can pass/>And/>The difference between the aggregate consistent value of the kth outlier cloud data and the average value of the aggregate consistent values of all outlier cloud data in the final cluster to which the kth outlier cloud data belongs is not limited herein.
And obtaining the matching degree value of each abnormal point cloud data according to the method for obtaining the matching degree value of the kth abnormal point cloud data.
(2) And acquiring coordinate points of the final defect point cloud data.
Preferably, the method for acquiring the coordinate points of the final defect point cloud data comprises the following steps: multiplying the matching degree value of each abnormal point cloud data with the corresponding coordinate point to obtain a corrected coordinate point corresponding to each abnormal point cloud data; knowing that all abnormal point cloud data in each final cluster are defaulted to point cloud data corresponding to the same defect point in the large-diameter penstock, the embodiment of the invention acquires the central coordinate point of the corrected coordinate point corresponding to all abnormal point cloud data in each final cluster as the coordinate point of each final defect point cloud data. The method for acquiring the center coordinate point is a known technology, and will not be described in detail. So far, the coordinate point of each final defect point cloud data is obtained.
(3) And detecting the quality of the large-diameter penstock.
And inputting coordinate points of all final defect point cloud data into a trained deep learning neural network model, and accurately outputting a defect region in the large-diameter penstock. The training process of the deep learning neural network model is as follows: in the embodiment of the invention, 1000 training is taken as an example, and an implementer can set training times according to actual conditions without limitation. The loss function of the deep learning neural network model is a cross entropy loss function. The deep learning neural network model is a known technology, and will not be described in detail. And then constructing a three-dimensional model of the large-diameter penstock and the defect area thereof, specifically evaluating the quality of the large-diameter penstock according to the size of the defect area displayed in the three-dimensional model, for example, when each defect area is smaller than a preset defect area threshold value, the quality of the large-diameter penstock is qualified by default, and when any defect area is larger than or equal to the preset defect area threshold value, the quality failure of the large-diameter penstock is determined, and an implementer can set the preset defect area threshold value according to actual conditions without limitation. Meanwhile, an operator can specifically evaluate the quality of the large-diameter pressure steel pipe according to actual conditions, the quality is not limited, and further accurate detection of the quality of the large-diameter pressure steel pipe is achieved by accurately acquiring a defect area in the large-diameter pressure steel pipe.
The present invention has been completed.
In summary, the embodiment of the invention obtains abnormal point cloud data of the large-diameter penstock; acquiring an aggregation degree value according to the distribution condition of abnormal point cloud data of the same position under the detection position and the distance between each abnormal point cloud data and adjacent abnormal point cloud data; acquiring an aggregation consistent value according to the quantity of the reference abnormal point cloud data of the abnormal point cloud data and the fluctuation condition of the aggregation degree value, further acquiring a neighborhood distance threshold value in a DBSCAN clustering algorithm, and clustering the abnormal point cloud data into a final cluster; and determining defects according to the aggregation consistent value and the coordinate point of the abnormal point cloud data in the final cluster, and detecting the quality of the large-diameter penstock. According to the invention, the defects in the large-diameter pressure steel pipe are determined by analyzing the distribution of the abnormal point cloud data, so that the quality of the large-diameter pressure steel pipe is accurately detected.
Based on the same inventive concept as the method embodiment, the embodiment of the invention also provides a large-diameter penstock quality detection system based on contour extraction, which comprises: the method comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to realize the steps in the embodiment of the large-diameter penstock quality detection method based on contour extraction, such as the steps shown in fig. 1. The method for detecting the quality of the large-diameter penstock based on contour extraction is described in detail in the above embodiments, and will not be described again.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.
Claims (10)
1. The large-diameter pressure steel pipe quality detection method based on contour extraction is characterized by comprising the following steps of:
acquiring abnormal point cloud data of the large-diameter pressure steel pipe at each detection position;
classifying the outlier point cloud data into point cloud data categories based on distances between outlier point cloud data; optionally selecting one piece of abnormal point cloud data as target point cloud data, and constructing the abnormal point cloud data of the same position as the target point cloud data in all detection positions as a point cloud data set; acquiring an aggregation degree value of the target point cloud data according to the distribution condition of the abnormal point cloud data in the point cloud data set in the point cloud data category to which the target point cloud data belong and the distance between the target point cloud data and the preset number of adjacent abnormal point cloud data;
Screening out reference abnormal point cloud data of the target point cloud data according to the distance between the target point cloud data and each other abnormal point cloud data in the class of the target point cloud data to which the target point cloud data belongs; acquiring an aggregation consistent value of the target point cloud data according to the quantity of the reference abnormal point cloud data and the fluctuation condition of the aggregation degree value;
Acquiring a neighborhood distance threshold value in a DBSCAN clustering algorithm according to the aggregation consistent value of each abnormal point cloud data, and clustering the abnormal point cloud data to acquire a final cluster; acquiring coordinate points of final defect point cloud data according to the aggregation consistent value and the coordinate points of each abnormal point cloud data in each final cluster, further determining defects in the large-diameter pressure steel pipe, and detecting the quality of the large-diameter pressure steel pipe;
the method for obtaining the defects in the large-diameter pressure steel pipe comprises the following steps: and inputting coordinate points of all final defect point cloud data into a trained deep learning neural network model, and accurately outputting defects in the large-diameter penstock.
2. The method for detecting the quality of the large-diameter penstock based on contour extraction as claimed in claim 1, wherein the method for acquiring the aggregation level value is as follows:
acquiring the number of abnormal point cloud data in the point cloud data set in the point cloud data category to which the target point cloud data belong as a first number of the target point cloud data;
And acquiring the aggregation degree value of the target point cloud data according to the first quantity, the ratio of the total quantity of all abnormal point cloud data in the point cloud data set to the first quantity and the Euclidean distance between the target point cloud data and the preset quantity of adjacent abnormal point cloud data.
3. The method for detecting the quality of the large-diameter penstock based on contour extraction as claimed in claim 2, wherein the calculation formula of the aggregation level value is as follows:
; in the/> The aggregation degree value of the kth abnormal point cloud data is obtained; /(I)The total number of all abnormal point cloud data in the point cloud data set where the kth abnormal point cloud data is located; /(I)A first quantity of kth outlier cloud data; m is a preset number; /(I)The Euclidean distance between the kth abnormal point cloud data and the coordinate point corresponding to the mth adjacent abnormal point cloud data; exp is an exponential function based on a natural constant.
4. The method for detecting the quality of the large-diameter penstock based on contour extraction as claimed in claim 1, wherein the method for acquiring the reference abnormal point cloud data is as follows:
acquiring Euclidean distances between the target point cloud data and coordinate points corresponding to other abnormal point cloud data in the class of the target point cloud data, wherein the target point cloud data belong to the target point cloud data, and the Euclidean distances are used as first distances;
and taking other abnormal point cloud data of the non-target point cloud data corresponding to the minimum first distance as reference abnormal point cloud data of the target point cloud data.
5. The method for detecting the quality of the large-diameter penstock based on contour extraction as claimed in claim 1, wherein the calculation formula of the aggregation consistent value is as follows:
; in the/> An aggregation consistent value for the kth outlier cloud data; /(I)The total number of reference outlier cloud data that is kth outlier cloud data; /(I)The aggregation degree value of the first reference abnormal point cloud data of the kth abnormal point cloud data; /(I)The average value of the aggregation degree values of all the reference abnormal point cloud data of the kth abnormal point cloud data; /(I)As a function of absolute value; norm is a normalization function; /(I)Is a first preset constant, greater than 0.
6. The method for detecting the quality of the large-diameter penstock based on contour extraction as claimed in claim 4, wherein the method for obtaining the neighborhood distance threshold value in the DBSCAN clustering algorithm according to the aggregation consistent value of each abnormal point cloud data is as follows:
Taking the abnormal point cloud data corresponding to the maximum aggregation consistent value as characteristic point cloud data;
And acquiring the average value of the first distances between all the characteristic point cloud data and the reference abnormal point cloud data thereof as a neighborhood distance threshold value in the DBSCAN clustering algorithm.
7. The method for detecting the quality of the large-diameter penstock based on contour extraction as claimed in claim 1, wherein the method for obtaining the coordinate points of the final defect point cloud data according to the aggregate consistent value and the coordinate points of each abnormal point cloud data in each final cluster is as follows:
Acquiring a matching degree value of each abnormal point cloud data according to the difference of the aggregation consistent value of each abnormal point cloud data and the average value of the aggregation consistent values of all the abnormal point cloud data in the final cluster to which the abnormal point cloud data belongs; wherein the difference and the matching degree value are in a negative correlation relationship;
multiplying the matching degree value of each abnormal point cloud data with the corresponding coordinate point to obtain a corrected coordinate point corresponding to each abnormal point cloud data;
And acquiring the central coordinate points of the corrected coordinate points corresponding to all the abnormal point cloud data in each final cluster, and taking the central coordinate points as the coordinate points of each final defect point cloud data.
8. The method for detecting the quality of the large-diameter penstock based on contour extraction as claimed in claim 1, wherein the method for acquiring the point cloud data category is as follows:
And acquiring Euclidean distance of coordinate points corresponding to any two abnormal point cloud data, and clustering the abnormal point cloud data by using an OPTICS density clustering algorithm to obtain the point cloud data category.
9. The method for detecting the quality of the large-diameter penstock based on contour extraction as claimed in claim 1, wherein the method for acquiring the abnormal point cloud data is as follows:
Acquiring point cloud data of the large-diameter pressure steel pipe at each detection position; wherein each point cloud data comprises ultrasound data and a reference distance;
and acquiring an ultrasonic data threshold value under each reference distance, and when the ultrasonic data of the point cloud data is larger than the ultrasonic data threshold value under the corresponding reference distance, the corresponding point cloud data is abnormal point cloud data.
10. A large diameter penstock quality inspection system based on profile extraction comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, performs the steps of a large diameter penstock quality inspection method based on profile extraction as claimed in any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410565741.8A CN118154588B (en) | 2024-05-09 | 2024-05-09 | Large-diameter pressure steel pipe quality detection method and system based on contour extraction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410565741.8A CN118154588B (en) | 2024-05-09 | 2024-05-09 | Large-diameter pressure steel pipe quality detection method and system based on contour extraction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118154588A true CN118154588A (en) | 2024-06-07 |
| CN118154588B CN118154588B (en) | 2024-07-19 |
Family
ID=91300225
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410565741.8A Active CN118154588B (en) | 2024-05-09 | 2024-05-09 | Large-diameter pressure steel pipe quality detection method and system based on contour extraction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118154588B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113487490A (en) * | 2021-05-24 | 2021-10-08 | 深圳亦芯智能视觉技术有限公司 | Method and device for detecting internal defects of pipeline through stereoscopic vision imaging |
| US20220092345A1 (en) * | 2020-09-23 | 2022-03-24 | Faro Technologies, Inc. | Detecting displacements and/or defects in a point cloud using cluster-based cloud-to-cloud comparison |
| CN114266891A (en) * | 2021-11-17 | 2022-04-01 | 京沪高速铁路股份有限公司 | Railway operation environment abnormity identification method based on image and laser data fusion |
| CN115272493A (en) * | 2022-09-20 | 2022-11-01 | 之江实验室 | Abnormal target detection method and device based on continuous time sequence point cloud superposition |
-
2024
- 2024-05-09 CN CN202410565741.8A patent/CN118154588B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20220092345A1 (en) * | 2020-09-23 | 2022-03-24 | Faro Technologies, Inc. | Detecting displacements and/or defects in a point cloud using cluster-based cloud-to-cloud comparison |
| CN113487490A (en) * | 2021-05-24 | 2021-10-08 | 深圳亦芯智能视觉技术有限公司 | Method and device for detecting internal defects of pipeline through stereoscopic vision imaging |
| CN114266891A (en) * | 2021-11-17 | 2022-04-01 | 京沪高速铁路股份有限公司 | Railway operation environment abnormity identification method based on image and laser data fusion |
| CN115272493A (en) * | 2022-09-20 | 2022-11-01 | 之江实验室 | Abnormal target detection method and device based on continuous time sequence point cloud superposition |
Non-Patent Citations (1)
| Title |
|---|
| 李屹;董思奇;史岩;陈琪: "基于激光点云数据的排水管道形变类缺陷定量检测技术", 《工程地球物理学报》, vol. 21, no. 1, 31 January 2024 (2024-01-31), pages 54 - 61 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118154588B (en) | 2024-07-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113689428B (en) | Mechanical part stress corrosion detection method and system based on image processing | |
| CN116186634B (en) | Intelligent management system for construction data of building engineering | |
| CN109115812A (en) | A kind of weld seam egative film defect identification method and system | |
| CN110992349A (en) | Underground pipeline abnormity automatic positioning and identification method based on deep learning | |
| CN115187527B (en) | Separation and identification method for multi-source mixed ultrahigh frequency partial discharge spectrum | |
| CN117556714B (en) | Preheating pipeline temperature data anomaly analysis method for aluminum metal smelting | |
| CN116541732B (en) | Meteorological monitoring system based on ultrasonic data and optimization algorithm | |
| CN116242443B (en) | Dynamic metering method of ultrasonic metering instrument | |
| CN111325738A (en) | Intelligent detection method and system for peripheral cracks of transverse hole | |
| CN117372433B (en) | Thickness parameter control method, device, equipment and storage medium | |
| CN115841488A (en) | Hole checking method of PCB (printed Circuit Board) based on computer vision | |
| CN112766301A (en) | Similarity judgment method for indicator diagram of oil extraction machine | |
| CN107908807B (en) | Small subsample reliability evaluation method based on Bayesian theory | |
| CN117251798A (en) | Meteorological equipment anomaly detection method based on two-layer progressive process | |
| CN110458809B (en) | Yarn evenness detection method based on sub-pixel edge detection | |
| CN114297264B (en) | Method and system for detecting abnormal fragments of time sequence signals | |
| US20230325413A1 (en) | Error Factor Estimation Device and Error Factor Estimation Method | |
| CN118154588B (en) | Large-diameter pressure steel pipe quality detection method and system based on contour extraction | |
| CN117237747A (en) | Hardware defect classification and identification method based on artificial intelligence | |
| US20230169681A1 (en) | Quantitative evaluation method and system for prediction result of remote sensing inversion | |
| CN116309493A (en) | Method and system for detecting defects of textile products | |
| CN109614758A (en) | The monitoring method of circular shape error with spatial coherence | |
| CN115546115A (en) | Ice-coated ice-shaped picture segmentation and identification method based on geographic morphology | |
| CN112037158B (en) | Shale gas field production equipment-based image enhancement labeling method | |
| Wang et al. | A High-accessibility Identification Technology for Butt Joint Weld Defects |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |