CN-117359155-B - Welding evaluation method, device and storage medium based on image data analysis

CN117359155BCN 117359155 BCN117359155 BCN 117359155BCN-117359155-B

Abstract

The invention relates to the technical field of welding, in particular to a welding evaluation method, a device and a storage medium based on image data analysis, wherein the method comprises the following steps of obtaining a video of a welding process to be evaluated; analyzing and extracting key information in the video, segmenting the video according to the key information and a preset welding flow segmentation standard, determining an operation stage corresponding to the video, determining evaluation information corresponding to the operation stage, determining the operation stage corresponding to the operation video according to the key information segmentation in the video by collecting operation video of a welding site, extracting the evaluation information in the determined operation stage, so that an evaluator evaluates the welding operation process according to the evaluation information, the requirements of the evaluator on mental concentration degree, reaction speed, evaluation proficiency degree and the like in the evaluation process are reduced, and the accuracy degree and the reliability of an evaluation result are improved.

Inventors

AN YONG
Du Xiaodai
ZHU YUNXIANG
LI ZHIYU

Assignees

北京洞微科技发展有限公司

Dates

Publication Date: 20260505
Application Date: 20220307

Claims (10)

1. A welding assessment method based on image data analysis, the method comprising the steps of: Acquiring a video of a welding process to be evaluated; Analyzing and extracting key information in the video; Dividing the video according to the key information and a preset welding flow dividing standard, and determining an operation stage corresponding to the video, wherein the operation stage is at least one of a preheating stage, a reaction stage, a calm stage, a pouring stage, a demolding stage and a tumor pushing stage; Determining evaluation information corresponding to the operation stage; The device comprises a preheating stage, a reaction stage, a calm stage, a pouring stage, a demolding stage and a tumor pushing stage, wherein the preheating stage, the reaction stage, the calm stage, the pouring stage, the demolding stage and the tumor pushing stage are respectively provided with a reference frame number, a process identification mark corresponding to each stage and a preset stage label threshold value corresponding to each stage; The video is segmented according to the key information and a preset welding flow segmentation standard, and the operation stage corresponding to the video is determined, which specifically comprises the following steps: extracting continuous images of a base reference frame number corresponding to an operation stage to be determined from the images of the video as judging images; And judging whether the total number of process identification marks corresponding to the operation stage to be determined in each judging image accords with a corresponding preset stage label threshold value.
2. The method as recited in claim 1, further comprising: storing and displaying the operation stage and the corresponding evaluation information; And/or the number of the groups of groups, Extracting a preset key information evaluation standard corresponding to the operation stage; comparing the evaluation information corresponding to the operation stage with a preset key information evaluation standard; and determining the welding quality of the welding process to be evaluated according to the comparison result.
3. The method according to claim 2, wherein the key information comprises at least the position and the size of a flask and flame, and wherein the analyzing and extracting the key information in the video comprises: Extracting an evaluation image from the video; Performing primary identification on the extracted evaluation image, and marking a primary identification result in the evaluation image; Determining a region of interest in the evaluation image according to the preliminary identification result; Processing the primary identification result in the region of interest to obtain a process identification mark; analyzing the process identification mark to identify the position and the size of the sand sealing box and the flame.
4. A method according to claim 3, wherein said determining a region of interest in said evaluation image based on said preliminary identification result comprises: Extracting a sand sealing box in the primary identification result; The method comprises the steps of determining an interested area by taking the width, the height and the position of the interested area as references, wherein the width of the left side edge of the interested area from the left side edge of the sand sealing box is 1-2 sand sealing box widths, the width of the right side boundary of the interested area from the right side edge of the sand sealing box is 1-2 sand sealing box widths, the height of the upper boundary of the interested area from the upper edge of the sand sealing box is 2.3-3.3 sand sealing box heights, and the height of the lower boundary of the interested area from the bottom edge of the sand sealing box is 0.5-1.5 sand sealing box heights.
5. The method according to claim 4, wherein the processing the preliminary identification result in the region of interest to obtain a process identification mark specifically includes: And cutting a part of the preliminary identification results outside the region of interest by taking a boundary frame of the region of interest as a frame, and reserving the preliminary identification result region in the region of interest and marking the preliminary identification result region as the process identification mark.
6. The method according to claim 5, wherein the segmenting the video according to the key information and a preset welding process segmentation standard, and determining the operation phase corresponding to the video specifically includes: And cutting the video according to a preset welding process, the position and the size of the sand sealing box in the video, the flame and the crucible to obtain at least one of a preheating stage, a reaction stage, a calm stage, a pouring stage, a demolding stage and a tumor pushing stage.
7. The method of claim 6, wherein the video is segmented according to a preset welding process, and the position and size of the sand sealing box, the flame and the crucible in the video, to obtain at least one of a preheating stage, a reaction stage, a calm stage, a pouring stage, a demolding stage and a tumor pushing stage, and the method specifically comprises: extracting continuous images of a base reference frame number corresponding to an operation stage to be determined from the images of the video as judging images; Judging whether the total number of process identification marks corresponding to the operation stage to be determined in each judging image accords with a corresponding preset stage label threshold value or not; When the total number of the process identification marks in the judging images is in a preset stage label threshold corresponding to an operation stage to be determined, updating the judging images frame by frame from front to back until the total number of the process identification marks in each judging image is larger than the preset stage label threshold for the first time, determining the time corresponding to the judging images as the starting time of the operation stage to be determined; Or alternatively And when the total number of the process identification marks in the judging images is in a preset stage label threshold corresponding to the operation stage to be determined, updating the judging images frame by frame from front to back until the total number of the process identification marks in each judging image is smaller than the preset stage label threshold last time, determining the time corresponding to the judging images as the ending time of the operation stage to be determined, and continuing updating the judging images frame by frame from back to front until the total number of the process identification marks in each judging image is larger than the preset stage label threshold last time, and determining the time corresponding to the judging images as the starting time of the operation stage to be determined.
8. The method of claim 7, wherein in the case that the operation phase is at least two of a preheating phase, a reaction phase, a calm phase, a casting phase, a demolding phase, and a tumor pushing phase; For two adjacent operation phases, when the ending time of the previous operation phase is later than the starting time of the next operation phase, the starting time earlier than the starting time of the next operation phase is determined, and the time corresponding to the group of the judging images closest to the starting time of the next operation phase is the ending time of the previous operation.
9. A welding assessment device based on image data analysis, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method according to any one of claims 1 to 8 when executing the program.
10. A computer readable storage medium storing a computer program, which when executed by a processor implements the method of any one of claims 1 to 8.

Description

Welding evaluation method, device and storage medium based on image data analysis The application is a divisional application of an application patent application with the application number 202210223473.2, wherein the application date is 2022, 3 and 7. Technical Field The invention relates to the technical field of welding, in particular to a welding evaluation method and device based on image data analysis and a storage medium. Background In recent years, with the development of the transportation industry, the demand for railway transportation is increasing, and the task of railway maintenance is becoming more important. For cracks and gaps of the steel rail, the maintenance and repair of the steel rail are generally carried out by adopting thermite welding and other technologies. The thermite welding process includes adhering sand mold to the steel rail to be welded to form two joints, preheating the two joints, setting thermite welding powder, ferric oxide and other alloy additives in certain proportion inside crucible to form thermite molten steel, igniting the thermite welding agent to produce redox reaction of ferric oxide, smelting the self-smelting plug in the bottom of the crucible, casting and filling the cavity to be welded, crystallizing the two ends of the rail, cooling to solidify to form welded joint, eliminating the sand mold, eliminating tumor and polishing. In the steel rail welding process, the operation is usually required to be carried out according to the regulations, and the carrying time or the carrying degree, the operating point position and the like of each operation stage such as a preheating stage, a reaction stage, a calm stage, a pouring stage, a demolding stage, a tumor pushing stage and the like are strictly controlled so as to avoid poor welding quality and even equipment damage. To ensure that the welding operation is in compliance, it is often necessary to evaluate each stage of the welding process separately. The current evaluation method is that an evaluation expert observes the welding process by naked eyes, and measures and calculates whether the proceeding time, the reaction degree and the like of each stage accord with the welding standard by experience and a stopwatch. For example, the preheating stage is that after flame is sprayed out of a pouring hole, an evaluation expert needs to immediately press a stopwatch to start preheating timing, the timing is immediately finished after flame is extinguished, the whole preheating process, particularly the process of changing color after rail web heating, is closely concerned in the preheating process, the temperature of the rail web after preheating is about 950-1000 ℃, the color of the rail web is bright yellow, the timing is started when molten iron flows out of a crucible in the pouring stage, and the pouring timing is finished when no molten iron flows out. Such a method for judging the operation process by the naked eyes of the evaluation expert has a problem that the demands for the concentration degree, the reaction speed, the evaluation proficiency and the like of the evaluation expert are high, the evaluation results are greatly affected by personal subjective factors of the evaluator, and there is a possibility that there is a large difference between the evaluation results of the respective evaluators. Disclosure of Invention The application aims to provide a welding evaluation method, a device and a storage medium based on image data analysis, which are used for acquiring operation videos of a welding site, determining operation stages corresponding to the operation videos according to key information in the videos, and extracting evaluation information in the determined operation stages so as to evaluate a welding operation process according to the evaluation information by an evaluator, thereby reducing the requirements on the mental concentration degree, the reaction speed, the evaluation proficiency degree and the like of the evaluator in the evaluation process and improving the accuracy and the reliability of an evaluation result. The application provides a welding evaluation method based on image data analysis, which comprises the following steps: The method comprises the steps of obtaining a video of a welding process to be evaluated, analyzing and extracting key information in the video, segmenting the video according to the key information and a preset welding flow segmentation standard, determining an operation stage corresponding to the video, and determining evaluation information corresponding to the operation stage. Further, the welding evaluation method provided by the invention further comprises the following steps: storing and displaying the operation stage and the corresponding evaluation information; And/or extracting a preset key information evaluation standard corresponding to the operation stage, comparing the evaluation information corresponding to the operation stage with the preset