CN-122023437-A - Power plant generator set design curve identification method and system

CN122023437ACN 122023437 ACN122023437 ACN 122023437ACN-122023437-A

Abstract

The invention relates to a method and a system for identifying a design curve of a generator set of a power plant, belongs to the technical field of curve image identification, and solves the problems of poor accuracy of existing curve identification and difficulty in multi-curve separation. The method comprises the steps of reading an original design curve image, preprocessing, dividing a candidate curve area through a communication component, screening an effective curve area from the candidate curve area to obtain a preliminarily divided curve image, extracting a plurality of curve frameworks, determining pixel points meeting preset conditions as starting points of the curve frameworks for each curve framework, starting from the determined starting points of each curve framework, searching and collecting all pixel point coordinates on the curve frameworks along the curve frameworks, generating independent curves based on all pixel point coordinates on the curve frameworks, converting the pixel point coordinates on each independent curve from image pixel coordinates to actual mathematical coordinates, and carrying out interpolation processing on the converted mathematical coordinates to generate a data table containing all independent curve mathematical coordinates.

Inventors

JIN SHENGXIANG
LI XIN
ZHAO JIANBO
DENG JIANPING
MA WANJUN
GONG ZHIPENG
CHEN HUILI
QIAN XIAOJUN
ZHANG SHIQUAN
ZHOU WEIXING

Assignees

北京京西燃气热电有限公司

Dates

Publication Date: 20260512
Application Date: 20251231

Claims (10)

1. The method for identifying the design curve of the power plant generator set is characterized by comprising the following steps of: reading an original design curve image, converting the original design curve image into a gray level image, and performing binarization processing to obtain a binary image; Detecting long straight lines in the binary image, creating a mask for the detected long straight lines, and removing a long straight line region marked by the mask from the binary image; Morphological operation is carried out on the image from which the long linear region is removed, candidate curve regions are segmented through the communication component, and effective curve regions are screened out from the candidate curve regions to obtain a preliminarily segmented curve image; skeletonizing the preliminarily segmented curve image to extract a plurality of curve skeletons; Traversing all pixel points on each curve skeleton respectively, and determining the pixel points meeting a preset condition as starting points of the curve skeleton for each curve skeleton, wherein the preset condition is that the starting points are minimum in abscissa values in eight adjacent domains and maximum in ordinate values when the abscissa values are the same, and the distance between the starting points is not smaller than a preset threshold value; For each curve skeleton, starting from the determined starting point, searching and collecting all pixel point coordinates on the curve skeleton along the curve skeleton, and generating an independent curve based on all pixel point coordinates on the curve skeleton; Converting pixel point coordinates on each independent curve from image pixel coordinates to actual mathematical coordinates; and carrying out interpolation processing on the converted mathematical coordinates to generate a data table containing all the mathematical coordinates of the independent curves.
2. The method of claim 1, wherein detecting long lines in the binary image and creating a mask for the detected long lines comprises: And identifying straight line segments in the binary image based on a straight line detection algorithm, screening out straight line segments with included angles smaller than a preset angle and lengths exceeding a preset length threshold value from the identified straight line segments, and creating mask marks for marking the areas where the straight line segments are located in the binary image by taking the screened straight line segments as references.
3. The method of claim 1, wherein morphologically manipulating the image after removing the long straight line regions comprises: Performing morphological open operation on the image from which the long linear region is removed to remove small-size discrete noise points; performing a morphological closing operation on the image from which the small-sized discrete noise points are removed to bridge the break in the curve; The effective curve area screening according to the area width comprises the steps of calculating the circumscribed rectangular width of each communication assembly and reserving the area where the communication assembly with the width larger than the preset minimum width value and smaller than the preset maximum width value is located.
4. The method of claim 1, wherein skeletonizing the preliminary segmented curve image to extract a plurality of curve skeletons comprises: performing a gaussian blur process on the preliminarily segmented graph image; performing binarization processing on the Gaussian blur processed image to obtain a clean binary image; And skeletonizing the clean binary image to obtain a curve skeleton image comprising a plurality of curve skeletons with single pixel widths.
5. The method of claim 1, wherein for each curve skeleton, starting from its determined starting point, searching and collecting all pixel point coordinates on the curve skeleton along the curve skeleton comprises: Constructing a stack data structure and a current curve pixel point coordinate set; Storing the starting point into the stack data structure, and marking the starting point as accessed; the following steps are circularly executed: step a, removing the pixel point at the stack top of the stack data structure, and taking the pixel point as a current point; B, adding the coordinates of the current point into the current curve pixel point coordinate set; step c, searching pixel points which are adjacent to the current point, belong to the curve skeleton and are not marked as accessed according to the preset direction priority; step d, marking the searched pixel points as accessed and sequentially storing the accessed pixel points in the stack data structure; and d, repeating the steps a-d until the stack data structure is empty, wherein the current curve pixel point coordinate set comprises all pixel point coordinates on the curve skeleton.
6. The method of claim 5, wherein the predetermined direction priorities are, in order from greater to lesser, right direction, lower right direction, upper right direction, lower direction, upper direction.
7. The method of claim 5, wherein generating an independent curve based on all pixel point coordinates on a curve skeleton comprises: counting the number of coordinate points in a current curve pixel point coordinate set corresponding to each curve skeleton; discarding the coordinate set of the pixel points of the current curve, the number of the coordinate points of which is less than a preset number threshold value; and respectively generating independent curves based on the rest coordinate sets of all the current curve pixel points.
8. The method of claim 1, wherein the pixel point coordinates on each independent curve are converted from image pixel coordinates to actual mathematical coordinates as shown in the following formula; x math ＝x pixel *(x max -x min /(w-1)+x min ; y math ＝(h-1-y pixel )*(y max -y min /(h-1)+y min ; Wherein x pixel is the image pixel abscissa, y pixel is the image pixel ordinate, x math is the converted actual mathematical coordinate abscissa, y math is the converted actual mathematical coordinate ordinate, h is the height of the original design curve image, w is the width of the original design curve image, x min is the preset mathematical coordinate system abscissa minimum, x max is the preset mathematical coordinate system abscissa maximum, y min is the preset mathematical coordinate system ordinate minimum, and y max is the preset mathematical coordinate system ordinate maximum.
9. The method of claim 1, wherein interpolating the converted mathematical coordinates comprises: Collecting mathematical coordinates of all the independent curves after conversion, determining the minimum value and the maximum value of the global abscissa, and generating a group of uniformly distributed full-scale abscissa points in the range determined by the minimum value and the maximum value; performing linear interpolation on the full-scale abscissa points for each independent curve to calculate a corresponding ordinate; a data table is created based on the full-scale abscissa points and their corresponding ordinate.
10. A power plant genset design curve identification system, comprising: The binarization processing module is used for reading an original design curve image, converting the original design curve image into a gray level image and then carrying out binarization processing to obtain a binary image; a removing module, configured to detect a long straight line in the binary image, create a mask for the detected long straight line, and remove a long straight line region marked by the mask from the binary image; the segmentation module is used for carrying out morphological operation on the image with the long linear region removed, segmenting a candidate curve region through the communication component, and screening out an effective curve region from the candidate curve region to obtain a primarily segmented curve image; the extraction module is used for carrying out skeletonization on the preliminarily segmented curve image to extract a plurality of curve skeletons; The starting point determining module is used for traversing all the pixel points on each curve skeleton respectively, determining the pixel points meeting the preset condition as the starting points of the curve skeleton for each curve skeleton, wherein the preset condition is that the starting points are minimum in the abscissa values in eight adjacent domains, and the ordinate values are maximum when the abscissa values are the same, and the distance between the starting points is not smaller than a preset threshold value; The searching module is used for tracking and collecting all pixel point coordinates on each curve skeleton from the determined starting point of each curve skeleton, and generating an independent curve based on all pixel point coordinates on the curve skeleton; the coordinate conversion module is used for converting the pixel point coordinates on each independent curve from the image pixel coordinates to actual mathematical coordinates; And the data table generation module is used for carrying out interpolation processing on the converted mathematical coordinates to generate a data table containing all the mathematical coordinates of the independent curves.

Description

Power plant generator set design curve identification method and system Technical Field The invention relates to the technical field of curve image recognition, in particular to a method and a system for recognizing a design curve of a generator set of a power plant. Background In the process of planning, running and maintaining a power plant generator set, a design curve (such as a performance curve, a starting curve and the like) is an important carrier for recording the change of key parameters of the generator set along with working conditions, and is usually stored in a drawing or image form. In order to use these curve data for digital analysis and application, it is necessary to convert them from an image format to accurate numerical information. Along with the improvement of the digitization and the intellectualization level of a power plant, the requirement of carrying out automation and high-precision identification on a design curve is increasingly urgent. At present, a common curve identification method faces a plurality of challenges when processing a power plant design curve image, such as complex drawing background, intersection and overlapping of a plurality of curves, interference of coordinate axes and labeling information and the like. The factors lead to the defect of insufficient accuracy of key links such as curve segmentation, skeleton extraction, multi-curve separation and the like in the prior art, the processing process depends on manual intervention, the efficiency is low, and the requirements of data precision and batch processing in engineering practice are difficult to meet. Disclosure of Invention In view of the above analysis, the embodiment of the invention aims to provide a method and a system for identifying a design curve of a generator set of a power plant, which are used for solving the problems of poor accuracy of identifying the existing curve and difficulty in separating multiple curves. In one aspect, an embodiment of the present invention provides a method for identifying a design curve of a generator set in a power plant, including: reading an original design curve image, converting the original design curve image into a gray level image, and performing binarization processing to obtain a binary image; Detecting long straight lines in the binary image, creating a mask for the detected long straight lines, and removing a long straight line region marked by the mask from the binary image; Morphological operation is carried out on the image from which the long linear region is removed, candidate curve regions are segmented through the communication component, and effective curve regions are screened out from the candidate curve regions to obtain a preliminarily segmented curve image; skeletonizing the preliminarily segmented curve image to extract a plurality of curve skeletons; Traversing all pixel points on each curve skeleton respectively, and determining the pixel points meeting a preset condition as starting points of the curve skeleton for each curve skeleton, wherein the preset condition is that the starting points are minimum in abscissa values in eight adjacent domains and maximum in ordinate values when the abscissa values are the same, and the distance between the starting points is not smaller than a preset threshold value; For each curve skeleton, starting from the determined starting point, searching and collecting all pixel point coordinates on the curve skeleton along the curve skeleton, and generating an independent curve based on all pixel point coordinates on the curve skeleton; Converting pixel point coordinates on each independent curve from image pixel coordinates to actual mathematical coordinates; and carrying out interpolation processing on the converted mathematical coordinates to generate a data table containing all the mathematical coordinates of the independent curves. Further, detecting a long straight line in the binary image, and creating a mask for the detected long straight line includes: And identifying straight line segments in the binary image based on a straight line detection algorithm, screening out straight line segments with included angles smaller than a preset angle and lengths exceeding a preset length threshold value from the identified straight line segments, and creating mask marks for marking the areas where the straight line segments are located in the binary image by taking the screened straight line segments as references. Further, the morphological operation on the image from which the long straight line region is removed includes: Performing morphological open operation on the image from which the long linear region is removed to remove small-size discrete noise points; performing a morphological closing operation on the image from which the small-sized discrete noise points are removed to bridge the break in the curve; The effective curve area screening according to the area width comprises the steps of calculati