WO2020228044A1

WO2020228044A1 - Multiscale abnormal region rapid recommendation system and method for pipeline magnetic flux leakage data

Info

Publication number: WO2020228044A1
Application number: PCT/CN2019/087346
Authority: WO
Inventors: 刘金海; 付明芮; 臧东; 卢森骧; 汪刚; 马大中; 徐可馨
Original assignee: 东北大学
Priority date: 2019-05-10
Filing date: 2019-05-17
Publication date: 2020-11-19
Also published as: CN110082424B; CN110082424A

Abstract

A multiscale abnormal region rapid recommendation system and method for pipeline magnetic flux leakage data, relating to the technical field of pipeline testing. The following steps are comprised: step 1, obtaining a magnetic flux leakage signal of a segment of pipeline, performing multiscale window division on the magnetic flux leakage signal, and performing abnormal edge extraction on N scale levels so as to obtain an abnormal window set; step 2, performing abnormal region estimation on the abnormal window set to obtain an abnormality estimation set; and step 3, boundary accuracy: calculating the area ratio of adjacent windows of all the windows in W^k" according to the abnormality estimation set, traversing the abnormality estimation set W', removing the window having the area ratio less than λ, and selecting an outermost peripheral window in the overlapping windows within the current set to be used as an abnormal recommendation region. By means of the method, the present invention can find an abnormality with large size and a clear signal, can also find a small abnormality, can provide a sufficiently abnormal candidate region, has significant rapidity, and is particularly applicable for a data set with a huge amount of pipelines.

Description

Fast pipeline magnetic flux leakage data multi-scale abnormal area recommendation system and method

Technical field

The invention relates to the technical field of pipeline detection, in particular to a system and method for recommending multi-scale abnormal regions of rapid pipeline magnetic flux leakage data.

Background technique

With its high efficiency, safety and reliability, pipeline transportation is called the five major transportation modes along with railway, highway, waterway and aviation. With the increase of pipeline service time, due to the influence of pipeline material problems, external damage and medium corrosion, the pipeline condition gradually deteriorates, and there is a potential risk of damage and leakage. Once a leak occurs, it will not only cause air pollution, but also a violent explosion. In 2011, an oil spill accident occurred in the Bohai Bay. According to statistics from the State Oceanic Administration, the accident leaked 385 cubic meters of crude oil, causing a total of 5,500 square kilometers of seawater pollution. Therefore, in order to ensure the safety of energy transportation and ecological environment, the pipeline must be regularly inspected and maintained.

Non-destructive testing (NDT) is widely used as an important means of pipeline safety maintenance. Among them, magnetic flux leakage detection as a non-destructive testing method is widely used in nearly 90% of in-service pipelines. A complete magnetic flux leakage data analysis process includes 5 parts, namely: data preprocessing, abnormal area recommendation, abnormal identification, defect size inverse estimation and defect safety assessment. The data preprocessing part completes the filtering of the base value correction of the original data; the abnormal area recommendation part obtains the position of the abnormal area; the abnormal recognition part completes the classification and recognition of the abnormal position, such as defects, valves, meters, etc.; the defect size estimation part realizes the defect signal To the size of the mapping, and the defect safety assessment part is to calculate the safety level of the defect to determine whether it needs repair.

The recommendation of abnormal areas is a key and challenging issue in the magnetic flux leakage data analysis process. A good anomaly area recommendation algorithm not only has position accuracy and edge accuracy, but also has the ability to be fast. In practical applications, the recommendation for abnormal regions is based on the traditional exhaustive search algorithm, without considering the influence of the sampling of candidate regions on the efficiency of the algorithm. The huge search space ultimately wastes a lot of time. At the same time, minor abnormalities are prone to missed detection due to noise.

Summary of the invention

The technical problem to be solved by the present invention is to provide a fast pipeline magnetic flux leakage data multi-scale abnormal area recommendation system and method for the above-mentioned shortcomings of the prior art. The present invention has obvious rapidity and is particularly suitable for huge pipeline data sets.

In order to solve the above technical problems, the technical solutions adopted by the present invention are:

On the one hand, the present invention provides a fast multi-scale abnormal area recommendation system for pipeline magnetic flux leakage data, including an input and output module, a multi-scale window division module, an abnormal area estimation module, and a boundary precision module;

The input and output module is used for inputting the magnetic leakage signal and outputting the abnormal target location area of the pipeline, and outputting the magnetic leakage signal to the multi-scale window division module;

The multi-scale window division module is used to complete the acquisition of the multi-scale candidate abnormal window, and output the abnormal window to the abnormal area estimation module;

The abnormal area estimation module is used to estimate the position of the abnormal area, obtain an abnormal estimation set, and output the set to the boundary precision module;

The boundary precision module is used to describe in detail the boundary of each window in the abnormality estimation set to obtain the abnormal recommendation area, and output the abnormal recommendation area to the input and output module after merging.

On the other hand, the present invention provides a method for recommending a multi-scale abnormal area of fast pipeline magnetic flux leakage data, which is implemented by the described system of recommending a multi-scale abnormal area of rapid pipeline magnetic flux leakage data, including the following steps:

Step 1: Obtain the magnetic flux leakage signal D of a section of pipeline, and divide it into a multi-scale window, and divide it into N scale levels L ₁ , L ₂ ,...L _N , and perform abnormal edge extraction on the N scale levels. Obtain the abnormal form set W={W ¹ , W ² ,..., W ^k ,...W ^N }; where,

Is the set of windows under the kth scale level, and b is the number of windows included in the kth scale level;

Step 2: Perform abnormal area estimation on the abnormal window set obtained in Step 1, and obtain an abnormal estimation set W″={W ^1″ , W ^2″ ,..., W ^k″ ,...W ^N″ };

Step 2.1: Preprocess the forms in the abnormal form collection, the preprocessing is to remove the unclosed borders, and obtain the processed abnormal form collection W′={W ^1′ , W ^2′ ,... , W ^k′ ,...W ^N′ };

Step 2.2: Estimate the score of the abnormal window set W'; for the abnormal area, there will be multiple windows stacked together to form multiple "back" glyphs, and each window corresponds to a measurement window intersection The overlap degree value S(W ^k′ ), traverse the abnormal window set W′, calculate the overlap degree value of each window, and the value is equivalent to the score of the abnormal window, the formula is as follows:

Among them, W ^k′ represents the current window, and m is the number of contour windows contained in the current window;

Respectively represent the number of windows in the kth window and the number of windows outside the kth window;

Step 2.3: Obtain the abnormal area, select the window with the score of the window greater than σ as the abnormal window, where 0≤σ≤1, and determine whether there is a parallel window at the same scale level, and if it exists, the scale will be lowered Delete all the side-by-side forms of, and get the abnormal estimation set W″={W ^1″ , W ^2″ ,..., W ^k″ ,...W ^N″ };

Step 3: The boundary is accurate; according to the abnormal estimation set W″={W ^1″ , W ^2″ ,..., W ^k″ ,…W ^N″ } obtained in step 2, all the forms in W ^k″ Carry out the area ratio of adjacent windows, traverse the abnormal estimation set W″, remove the windows whose area ratio is less than λ, where 0≤λ≤1, and select the outermost window in the overlapping windows in the current set as the abnormal recommendation area ；

The formula for area ratio is as follows:

among them,

Is the area ratio of the h-1th window to the hth window under the kth scale level; ||*|| represents the window area, that is, the number of data points contained in the window;

Respectively represent the h-1th window and the hth window under the kth scale level;

The specific steps of step 1 are as follows:

Step 1.1: For a section of pipeline magnetic flux leakage signal D, divide the minimum and maximum values of the signal into N scale levels L ₁ , L ₂ ,...L _N , then the value of the k-th scale level is:

Step 1.2: For the scale level k, slice the received value L _{k of} the scale level and the current signal D to obtain a binary matrix D ^k , namely:

Where D _i,j are the data points in the i-th row and j-th column in the pipeline magnetic flux leakage signal D;

Step 1.3: Perform abnormal edge extraction on the binary matrix D ^k .

Step 1.3.1: Establish two templates in orthogonal directions: horizontal template fx and vertical template fy, namely: fy=[-11]; fx=fy ^T , fx represents the transposition of fy.

Step 1.3.2: Use the above horizontal template fx and vertical template fy to filter the binary matrix D ^k in two directions respectively to obtain the filtered binary matrix D ^k,x and D ^k,y ; the binary matrix The edge matrix of D ^k is E ^k :

Step 1.3.3: Obtain the abnormal edges, and then regularize the abnormal edges to form a rectangular window to obtain the current scale set abnormal window set W ^k ;

Step 1.4: Repeat step 1 to step 3 to obtain the abnormal window set W={W ¹ , W ² ,..., W ^k ,...W ^N } of all scale levels.

The beneficial effect produced by the above technical solution is that the present invention provides a fast pipeline magnetic flux leakage data multi-scale abnormal area recommendation system and method. By dividing the magnetic flux leakage data in multiple scales, the multi-scale window combination is obtained. By deleting and merging the series of the form, the abnormal candidate area is finally obtained. Starting from a multi-scale perspective, the present invention extracts data at multiple data levels. The extraction process embodies from coarse to fine. Compared with general anomaly extraction algorithms, the present invention can not only find abnormalities with larger size and obvious signals. At the same time, small abnormalities can be found, and sufficient abnormal candidate regions can be provided; the present invention deals with multi-scale windows, expresses signal characteristics in windows, and finally determines the target. Compared with general abnormalities based entirely on signal Extraction algorithm, the invention has obvious rapidity, especially suitable for the huge data set of the pipeline.

Description of the drawings

FIG. 1 is a block diagram of a system for recommending a multi-scale abnormal area of fast pipeline magnetic flux leakage data according to an embodiment of the present invention;

2 is a flowchart of a method for recommending a multi-scale abnormal area of fast pipeline magnetic flux leakage data according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of magnetic flux leakage data provided by an embodiment of the present invention;

4 is a schematic diagram of multi-scale slice division provided by an embodiment of the present invention;

FIG. 5 is a schematic diagram of a multi-scale abnormal edge provided by an embodiment of the present invention;

6 is a schematic diagram of multi-scale abnormal edge regularization provided by an embodiment of the present invention;

FIG. 7 is a schematic diagram of a score estimation window provided by an embodiment of the present invention;

FIG. 8 is a schematic diagram of a maximum containment relationship combination window provided by an embodiment of the present invention;

FIG. 9 is a schematic diagram of an accurate boundary window provided by an embodiment of the present invention;

10 is a schematic diagram of the largest peripheral window provided by an embodiment of the present invention;

FIG. 11 is a schematic diagram of a defect signal target area provided by an embodiment of the present invention.

Detailed ways

The specific embodiments of the present invention will be described in further detail below in conjunction with the drawings and embodiments. The following examples are used to illustrate the present invention, but not to limit the scope of the present invention.

This embodiment is described below.

On the one hand, the present invention provides a fast multi-scale abnormal area recommendation system for pipeline magnetic flux leakage data, as shown in FIG. 1, including an input and output module, a multi-scale window division module, an abnormal area estimation module, and a boundary precision module;

The multi-scale window division module is used to complete the acquisition of the multi-scale candidate abnormal window and output the abnormal window to the abnormal area estimation module;

The abnormal area estimation module is used to estimate the position of the abnormal area, obtain an abnormal estimation set, and output the set to the boundary precision module.

On the other hand, the present invention provides a fast pipeline magnetic flux leakage data multi-scale abnormal area recommendation method, which is implemented by the fast pipeline magnetic flux leakage data multi-scale abnormal area recommendation system, as shown in Fig. 2, including the following steps:

Step 1: Obtain the magnetic flux leakage signal D of a section of pipeline, and divide it into a multi-scale window, and divide it into N scale levels L ₁ , L ₂ ,...L _N , and perform abnormal edge extraction on the N scale levels. Obtain the abnormal window set W={W ¹ , W ² ,..., W ^k ,...W ^N }. among them,

In this embodiment, N=40;

Step 1.1: For a section of pipeline magnetic flux leakage signal D, as shown in Figure 3, divide the minimum and maximum values of the signal into N scale levels L ₁ , L ₂ ,...L _N , then the k-th scale level The numerical size of is:

In this embodiment, the pipeline magnetic flux leakage signal D is divided into 40 scale levels;

Step 1.2: For the scale level k, the scale level collected value L _k and the current signal D are sliced, as shown in Figure 4, to obtain a binary matrix D ^k , namely:

Step 1.3: Perform abnormal edge extraction on the binary matrix D ^k .

Step 1.3.1: Establish two templates in orthogonal directions: horizontal template fx and vertical template fy, namely: fy=[-1 1]; fx=fy ^T , fx represents the transposition of fy.

Step 1.3.3: Obtain abnormal edges, as shown in Figure 5, and then regularize the abnormal edges to form a rectangular window, as shown in Figure 6, to obtain the current scale set abnormal window set W ^k .

Step 2: Perform abnormal area estimation on the abnormal window set obtained in Step 1, and obtain the abnormal estimation set W″={W ^1″ , W ^2″ ,..., W ^k″ ,...W ^N″ }.

Step 2.1: Preprocess the forms in the abnormal form collection, the preprocessing is to remove the unclosed borders, and obtain the processed abnormal form collection W′={W ^1′ , W ^2′ ,... , W ^k′ ,...W ^N′ }; we define that the necessary and sufficient condition for the existence of an abnormal window is: abnormal edge points can form a closed area connected end to end.

Step 2.2: Estimate the score of the abnormal window set W′; in practice, after the source data is divided into multiple scales and levels, for the abnormal area, there will be multiple windows stacked together to form multiple "backs" Font. Therefore, in order to characterize this feature, each window corresponds to a value S(W ^k′ ) that measures the degree of window overlap, traverse the abnormal window set W′, and calculate the value of the window overlap degree of each window , The value is equivalent to the score of the abnormal form, the formula is as follows:

Respectively represent the number of windows in the kth window and the number of windows outside the kth window.

Step 2.3: Obtain the abnormal area, select the window with the score of the window greater than σ as the abnormal window, where 0≤σ≤1, as shown in Figure 7, and judge whether there are parallel windows at the same scale level. If it exists, delete all the side-by-side windows under the scale, as shown in Figure 8, to obtain an abnormal estimation set W″={W ^1″ , W ^2″ ,..., W ^k″ ,...W ^N″ };

In this embodiment, σ=0.7;

Select the largest containment relationship combination form to remove the interference form, that is: if there is a form inside each form, the internal form must belong to a complete containment relationship.

Step 3: The boundary is accurate; as shown in Figure 9, according to the abnormal estimation set W″={W ^1″ , W ^2″ ,..., W ^k″ , …W ^N″ } obtained in step 2, the set after each of the above steps for any of the goal area, will cover more than one form, in order to obtain accurate border, we offer a spatial correlation ratio, to measure than the adjacent area of the form, the W ^{k "in} Perform the area ratio of adjacent windows to all the windows of, traverse the abnormal estimation set W″, remove windows with an area ratio less than λ, where 0≤λ≤1, and λ=0.5 in this embodiment; and select the removal area ratio to be less than The outermost window in the overlapping window in the set behind the window of λ is used as the abnormal recommended area, as shown in Figure 10;

The formula for area ratio is as follows:

among them,

In this embodiment, it is a magnetic leakage signal defect. The magnetic leakage signal defect will generate three magnetic leakage signal area windows. The middle window corresponds to the peak position, and the two windows correspond to the valley position. The three windows are combined to obtain The final target location area is shown in Figure 11;

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: The technical solutions recorded in the foregoing embodiments are modified, or some or all of the technical features thereof are equivalently replaced; these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the claims of the present invention.

Claims

A fast pipeline magnetic flux leakage data multi-scale abnormal area recommendation system, which is characterized in that it includes an input and output module, a multi-scale window division module, an abnormal area estimation module, and a boundary precision module;

The input and output module is used for inputting the magnetic leakage signal and outputting the abnormal target location area of the pipeline, and outputting the magnetic leakage signal to the multi-scale window division module;

The multi-scale window division module is used to complete the acquisition of the multi-scale candidate abnormal window, and output the abnormal window to the abnormal area estimation module;

The abnormal area estimation module is used to estimate the position of the abnormal area, obtain an abnormal estimation set, and output the set to the boundary precision module;

The boundary precision module is used to describe in detail the boundary of each window in the abnormality estimation set to obtain the abnormal recommendation area, and output the abnormal recommendation area to the input and output module after merging.
A fast multi-scale abnormal area recommendation method for pipeline magnetic flux leakage data, implemented by the fast multi-scale abnormal area recommendation system for pipeline magnetic flux leakage data according to claim 1, characterized in that it comprises the following steps:

Step 1: Obtain the magnetic flux leakage signal D of a section of pipeline, and divide it into a multi-scale window, and divide it into N scale levels L 1 , L 2 ,...L N , and perform abnormal edge extraction on the N scale levels. Obtain the abnormal form set W={W 1 , W 2 ,..., W k ,...W N }; where,
Is the set of windows under the kth scale level, and b is the number of windows contained in the kth scale level;

Step 2: Perform abnormal area estimation on the abnormal window set obtained in Step 1, and obtain an abnormal estimation set W″={W 1″ , W 2″ ,..., W k″ ,...W N″ };

Step 2.1: Preprocess the forms in the abnormal form collection, the preprocessing is to remove the unclosed borders, and obtain the processed abnormal form collection W′={W 1′ , W 2′ ,... , W k′ ,...W N′ };

Step 2.2: Estimate the score of the abnormal window set W'; for the abnormal area, there are multiple windows stacked together to form multiple "back" glyphs, and each window corresponds to a measurement window overlap The value of the degree S(W k′ ), traverse the abnormal window set W′, calculate the overlap degree value of each window, and the value is equivalent to the score of the abnormal window, the formula is as follows:

Among them, W k′ represents the current window, and m is the number of contour windows contained in the current window;
Respectively represent the number of windows in the kth window and the number of windows outside the kth window;

Step 2.3: Obtain the abnormal area, select the window with the score of the window greater than σ as the abnormal window, where 0≤σ≤1, and determine whether there is a parallel window at the same scale level, and if it exists, the scale will be lowered Delete all the side-by-side forms of, and get the abnormal estimation set W″={W 1″ , W 2″ ,..., W k″ ,...W N″ };

Step 3: The boundary is accurate; according to the abnormal estimation set W″={W 1″ , W 2″ ,..., W k″ ,…W N″ } obtained in step 2, all the forms in W k″ Carry out the area ratio of adjacent windows, traverse the abnormal estimation set W″, remove the windows with area ratio less than λ, where 0≤λ≤1, and select the outermost window in the overlapping windows in the current set as the abnormal recommendation area ；

The formula for area ratio is as follows:

among them,
Is the area ratio of the h-1th window to the hth window under the kth scale level; ||*|| represents the window area, that is, the number of data points contained in the window;
They respectively represent the h-1th window and the hth window under the kth scale level.
The method for recommending multi-scale abnormal regions of fast pipeline magnetic flux leakage data according to claim 2, wherein the specific steps of step 1 are as follows:

Step 1.1: For a section of pipeline magnetic flux leakage signal D, divide the minimum and maximum values of the signal into N scale levels L 1 , L 2 ,...L N , then the value of the k-th scale level is:

Step 1.2: For the scale level k, slice the received value L k of the scale level and the current signal D to obtain a binary matrix D k , namely:

Where D i,j are the data points in the i-th row and j-th column in the pipeline magnetic flux leakage signal D;

Step 1.3: Perform abnormal edge extraction on the binary matrix D k ;

Step 1.3.1: Establish two templates in orthogonal directions: horizontal template fx and vertical template fy, namely: fy=[-1 1]; fx=fy T , fx represents the transposition of fy;

Step 1.3.2: Use the above horizontal template fx and vertical template fy to filter the binary matrix D k in two directions respectively to obtain the filtered binary matrix D k,x and D k,y ; the binary matrix The edge matrix of D k is E k :

Step 1.3.3: Obtain the abnormal edges, and then regularize the abnormal edges to form a rectangular window to obtain the current scale set abnormal window set W k ;

Step 1.4: Repeat step 1 to step 3 to obtain the abnormal window set W={W 1 , W 2 ,..., W k ,...W N } of all scale levels.