High-speed image analysis method for interaction structure among multiple jet flames

Abstract

The invention provides a high-speed image analysis method suitable for identifying an interaction structure between flames through multi-jet-flow flames, and belongs to the technical field of flame structure analysis. According to the invention, flame self-luminous / induced fluorescence images shot by a high-speed camera are processed through a digital image processing method, so that the overlapping areas among the multi-jet flames are identified; the method specifically comprises the following steps of analysis area selection, analysis area segmentation, Gaussian filtering, feature corner point detection, feature corner point supplementation, optical flow field calculation, flame area discrimination, flame outer contour recognition and adjacent flame interaction extraction. According to the high-speed image analysis method suitable for identifying the interaction structure between the flames by the multi-jet-flow flames, the interaction structure between the flames can be identified from the flame image, the identification efficiency is high, a method can be provided for further analyzing the interaction between the flames, and the method is suitable for popularization and application. The device is very suitable for being used in the field of flame structure research in a multi-nozzle combustion chamber.

Description

Technical Field

[0001] The invention relates to the technical field of flame structure analysis, and in particular to a high-speed image analysis method for interaction structures among multiple jet flames. Background Art

[0002] In modern combustion technology, with the continuous improvement of energy efficiency and environmental protection requirements, the optimization of the combustion process has become a key research field. Multi-injection nozzle arrangements are often used in various combustion chambers, such as rocket engines, afterburners, and micro-burners. In a multi-injection nozzle combustion chamber, multiple flames coexist in a limited space, and the flames interact with each other, presenting complex flame characteristics. The temperature and heat release rate distribution in the combustion chamber are affected, thereby changing the combustion efficiency, pollutant emissions, and combustion stability of the combustion system. Therefore, analyzing and understanding the interaction structure between multiple flames is crucial to the optimization of the combustion process.

[0003] Experimental research is an important means in the field of flame structure research. In the experiment, high-speed cameras can capture the shape, boundary, brightness distribution and other information of the flame in real time, and accurately analyze the interaction structure of the flame. Especially in the case of multiple flames, the flames generated by different nozzles affect each other, which may lead to flame merging, squeezing, and even unstable combustion. These phenomena can be directly reflected at the visual level, and then combined with advanced image processing and data analysis techniques, the key features of multiple flames can be extracted. This information is of great significance for understanding the interaction mechanism of multiple jet flames and predicting their impact on combustion performance. Existing analysis methods mainly focus on the study of single flames or simple jet flame structures, and the comprehensive analysis of the interaction of multiple jet flames is relatively lacking.

[0004] Therefore, there is an urgent need for a visual analysis method specifically for the interaction structure of multiple jet flames, which can more accurately identify and track multiple flame boundaries, analyze the interaction behavior of flames in real time, and provide a basis for the design and optimization of combustion systems. Summary of the invention

[0005] To solve the above problems, the present invention provides a high-speed image analysis method for the interaction structure between multiple jet flames, which can capture the interaction structure between multiple jet flames from the flame image, and is particularly suitable for the extraction of the interaction structure between radially distributed multiple flames represented by afterburner flames.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A high-speed image analysis method for the interaction structure between multiple jet flames comprises the following steps:

[0008] S1. Image acquisition: outside the fuel injection port of the multi-injection port combustion chamber, a high-speed video of multi-jet flame self-luminescence is obtained by using a high-speed camera, and the high-speed video is processed by an image interface program to obtain a plurality of single-frame images, wherein the flame self-luminescence includes methyl self-luminescence and hydroxyl self-luminescence;

[0009] S2. Analysis area selection: selecting the analysis area for the single frame image, excluding the area upstream of the fuel injection port during the selection process;

[0010] S3, initial division of the analysis area: according to the spatial position of each fuel injection port, the analysis area is initially divided, and the center lines of the two injection ports are used as a reference, and the upper and lower center lines and the upstream and downstream boundaries of a certain injection port are surrounded to form an initial rectangular analysis area;

[0011] S4, Gaussian filtering: using Gaussian filtering to smooth the initial rectangular analysis area to reduce image noise;

[0012] S5, feature corner point detection: perform feature corner point detection on the current frame image after Gaussian filtering to obtain a special corner point set that can represent key feature points in the flame flow process;

[0013] S6, feature corner point supplement: repeat the operation of step S5 to process the next frame of image, obtain a feature corner point set, and compare it with the special corner point set of the previous frame of image, identify the feature corner points that are far away from the feature corner points of the previous frame, and save these feature corner points to the corner point set of the current frame;

[0014] S7, optical flow field calculation: match the feature corner points in two adjacent frames, track the displacement of each feature corner point in the continuous frames, and obtain the motion vector set of each feature point in the flame;

[0015] S8, flame region identification: analyzing the acquired motion vector set, judging and classifying the flame structure according to the motion direction and speed, judging the flame region to which each characteristic corner point belongs, so that the characteristic corner points in the current field form flame point clouds of different flame sources;

[0016] S9, flame outer contour recognition: performing concave hull outer contour recognition on each flame point cloud to extract the outer contour shape of the flame;

[0017] S10, adjacent flame interaction extraction: perform bitwise AND operation on each flame point cloud to obtain the interaction area between adjacent flames, and finally obtain the interaction structure between multiple jet flames.

[0018] Furthermore, in step S5, the goodFeatureToTrack algorithm is used to perform feature corner point detection on the current frame image after Gaussian filtering.

[0019] Furthermore, in step S7, the Lucas-Kanade pyramid iteration algorithm is used to match the feature corner points in two adjacent image frames.

[0020] Furthermore, in step S8, when determining the flame region to which each characteristic corner point belongs, if it is the first determination, the determination is made according to the rectangular space where the characteristic corner point is located; if it is a subsequent determination, the determination is made according to the flame source to which the upstream characteristic corner point of the vector belongs.

[0021] Furthermore, in step S9, the Moreira-Santos algorithm based on K-means clustering is used to perform flame contour recognition.

[0022] The present invention also provides an application of the above-mentioned high-speed image analysis method for the interaction structure between multiple jet flames, and the high-speed image analysis method is used to extract and analyze the interaction structure between radially distributed multiple flames.

[0023] In some embodiments, the radially distributed multiple flames are afterburner flames.

[0024] In some embodiments, the radially distributed multiple flames are rocket engine flames.

[0025] In some embodiments, the radially distributed multiple flames are multi-injection port combustion chamber flames.

[0026] The beneficial effects of the present invention are:

[0027] (1) The present invention discloses a high-speed image analysis method for the interaction structure between multiple jet flames. By processing the high-speed video of the jet flame group, the flow contour of each jet flame and the two-dimensional structure of the interaction interval can be captured in real time and dynamically, revealing the mutual interference and merging of flames, helping researchers to better understand the behavior of flames under different working conditions and optimize the control strategy in the combustion process.

[0028] (2) The present invention discloses a high-speed image analysis method for the interaction structure between multiple jet flames. The Moreira-Santos algorithm based on K-means clustering can identify the outer contour of the flame point cloud and effectively reproduce the complex curved structure of the flame edge, especially in the interaction area of ​​multiple jet flames. It provides clear spatial distribution information of the interaction between flames, helping researchers to more intuitively understand the impact of multiple flame interactions on combustion performance.

[0029] (3) The present invention discloses a high-speed image analysis method for the interaction structure between multiple jet flames, which provides a technical tool for flame interaction research and has high recognition efficiency. It can be widely used in flame structure analysis of complex combustion systems such as afterburners, rocket engines, and multi-jet combustion chambers.

[0030] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and implement it according to the contents of the specification, the following is a detailed description of the preferred embodiments of the present invention in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, some of the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0032] Figure 1 It is a schematic diagram of the analysis method shown in the present invention;

[0033] Figure 2 It is a schematic diagram of the combustion chamber structure shown in a preferred embodiment of the present invention;

[0034] Figure 3 It is a schematic diagram of a high-speed image of a multi-jet flame methyl self-luminescence shown in a preferred embodiment of the present invention;

[0035] Figure 4 is a schematic diagram of analysis area selection shown in a preferred embodiment of the present invention;

[0036] Figure 5 It is a schematic diagram of the initial segmentation of the analysis area shown in a preferred embodiment of the present invention;

[0037] Figure 6 is a schematic diagram of Gaussian filtering shown in a preferred embodiment of the present invention;

[0038] Figure 7 It is a schematic diagram of two-dimensional structure extraction of the interaction area between multiple flames shown in a preferred embodiment of the present invention.

[0039] Among them, the labels of various components are:

[0040] 1. Combustion chamber air inlet; 2. Combustion chamber fuel inlet; 3. Blunt body structure; 4. Combustion chamber outlet; 5. Motion vector of characteristic corner points; 6. Flame outer contour; 7. Flame interaction structure. DETAILED DESCRIPTION

[0041] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0042] The technical solutions described in the present invention, unless otherwise specified, are all conventional solutions in the art; the reagents or materials described, unless otherwise specified, are all from commercial channels.

[0043] Please refer to the present invention, which discloses a high-speed image analysis method for the interaction structure between multiple jet flames. The method uses a digital image processing method to process the flame methyl or hydroxyl self-luminous image for multiple jet flames, and extracts the flow contour of each jet flame and the two-dimensional structure of the interaction interval. The method includes the following steps:

[0044] S1. Image acquisition: This embodiment is based on the blunt body flame in the afterburner. Figure 2 This is a schematic diagram of the combustion chamber structure of the embodiment. The bluff body and the injection port are designed in an integrated manner. On the two sides of the bluff body, the symmetry axis in the height direction of the bluff body is the center line, and 4 rows of injection ports are evenly distributed on both sides. The injection port aperture is 1.5mm, and the injection port spacing is 8mm. The fuel from each row of injection ports is mixed with the air and burns at the trailing edge of the bluff body. The 4 flames have mutual influence. In the embodiment, propane fuel is used, and a multi-flame perspective video is obtained by a high-speed camera combined with a methyl filter. Each frame of the image is obtained through an image interface program. The image resolution is 1280×360 and the sampling rate is 5000Hz. Figure 3 It is a schematic diagram of a multi-flame image.

[0045] S2. Analysis area selection: First, select the analysis area of ​​interest for the image. Manual selection of the appropriate area is allowed. All pixels outside the area of ​​interest will be cut off. It should be noted that the area upstream of the fuel injection port should be excluded as much as possible to reduce the impact of pixel changes in the upstream area on the analysis results. Figure 4 is a schematic diagram of the selected analysis area.

[0046] S3. Initial segmentation of the analysis area: In this embodiment, there are 4 rows of fuel injection ports, and 4 flames are presented on the image. In the selected analysis area, a preliminary rectangular analysis area is constructed around each injection port. Specifically, the center lines of the two injection ports are used as a reference, and the upper and lower center lines and the upstream and downstream boundaries of a certain injection port are surrounded to segment the rectangular area of ​​each analysis flame. In this way, the combustion areas of the 4 flames are separated to ensure that the initial outline of each flame can be accurately identified, helping the subsequent image processing algorithm to better locate the boundaries of the flames and avoid blurring of the outlines caused by the interweaving of flames.

[0047] S4, Gaussian filtering: Preprocess the image of the selected analysis area and use the Gaussian filtering algorithm to smooth the image. The Gaussian filter can effectively reduce the noise in the image, especially in the high-speed camera process. Due to the limitations of shooting conditions, the flame image may be affected by noise. Through Gaussian filtering, the subtle changes of the flame edge can be smoothed to avoid noise interference affecting the subsequent analysis results. The Gaussian function formula is as follows:

[0048] ;

[0049] in, is the Gaussian function at position The weight of is the standard deviation, It is the coordinate offset with the center of the kernel as the reference point, indicating the distance from the center of the filter.

[0050] S5. Feature corner detection: Use the GoodFeatureToTrack algorithm to detect feature corners. This algorithm analyzes the grayscale value changes of the image and searches for points with significant directional changes in the local area. Specifically, it mainly relies on the gradient matrix of the image. and its eigenvalues, and the corner points in the image are determined by the minimum eigenvalue, such as (2) and (3). This algorithm can effectively identify the corner features in the image, and these corners can represent the key feature points in the flame flow process. In this step, the feature points in the current frame image are first detected, and a reasonable threshold is set to ensure that enough feature points can be detected. Specifically, the feature points should be distributed in the key area of ​​the flame, especially the edge area of ​​the flame, to ensure that the flow of the flame can be accurately identified in subsequent tracking.

[0051] ;

[0052] in, is the Gaussian window function, and The images are and Direction of grayscale gradient.

[0053] ;

[0054] if If is large, it means that the gray value of the point has significant changes in both directions, and it is a corner point. If the value is smaller, the point may be an edge or a flat area.

[0055] S6, feature corner point supplement: similarly perform feature corner point detection on the flame image of the next frame to obtain a feature corner point set, and by comparing it with the feature corner point set of the flame image of the previous frame, identify those feature corner points that are farther away from the corner points of the previous frame. Most of these points represent new flame structures transmitted from the upstream area during the flame flow process. Save these new feature corner points to the corner point set of the current frame.

[0056] S7. Optical flow calculation: Use the Lucas-Kanade pyramid iteration algorithm to match the feature corner points in two adjacent frames and track the displacement of each feature corner point in the continuous frames. This algorithm is based on the optical flow equation, such as (4), and can effectively track the motion trajectory of the feature points and obtain the motion vector set of each feature point in the flame. By tracking the feature corner points, the flow direction, speed and position changes of the flame can be recorded.

[0057] ;

[0058] in, It is the grayscale change of the image in the time dimension, that is, the difference between two frames. and is the optical flow field to be estimated, that is, the pixels in and Direction displacement.

[0059] S8, flame area identification: Analyze the acquired motion vector set, and judge and classify the flame structure according to the motion direction and speed. Determine the flame area to which each characteristic corner point belongs. If it is the first identification, it is determined according to the rectangular space where the point is located; if it is a subsequent identification, it is determined according to the flame source to which the upstream point of the vector belongs. In this way, the characteristic corner points in the current field form point clouds of different flame sources.

[0060] S9. Flame contour recognition: Since the flame structure is long and narrow, the Moreira-Santos algorithm based on K-means clustering is used to identify the concave hull contour of each flame point cloud. K-means clustering divides the data points into k clusters, and each data point is assigned to the cluster center closest to it, as shown in (5). The Moreira-Santos algorithm introduces additional constraints to ensure that the identified cluster center is more consistent with the actual boundary of the flame. During the flow of the flame, the edge is usually irregular and exhibits complex deformation. This algorithm can effectively extract the outer contour shape of the flame and reproduce the flow trend of the flame.

[0061] ;

[0062] in, It is the objective function, which is used to measure the quality of clustering results. is the number of clusters. It is Clusters. It is data points. It is The center point of a cluster.

[0063] S10, adjacent flame interaction extraction: By performing bitwise AND operations on each flame point cloud, the interaction area between adjacent flames can be obtained. These areas are usually the mutual interference intervals of flames during the combustion process, which can reflect the actual situation of the mutual influence of multiple jet flames during the blunt body combustion process.

[0064] This embodiment uses a digital image processing algorithm to effectively analyze the image of a blunt-body multi-flame methyl self-luminous flame with an afterburner as the background, and can effectively extract the two-dimensional structure of each flame interaction interval visually, which is conducive to further analysis and research on the mechanism of interaction between flames and their impact on combustion.

[0065] In summary, the present invention discloses a high-speed image analysis method for the interaction structure between multiple jet flames. By processing the high-speed video of the jet flame group, the flow contour of each jet flame and the two-dimensional structure of the interaction interval can be captured in real time and dynamically, revealing the mutual interference and merging between flames, helping researchers to better understand the behavior of flames under different working conditions and optimize the control strategy in the combustion process; the method is based on the Moreira-Santos algorithm of K-means clustering, which identifies the outer contour of the flame point cloud and effectively reproduces the complex curved structure of the flame edge, especially in the interaction area of ​​multiple jet flames, providing clear spatial distribution information of the interaction between flames, helping researchers to more intuitively understand the influence of the interaction between multiple flames on the combustion performance; the method provides a technical tool for the study of flame interaction, and has high recognition efficiency, and can be widely used in the flame structure analysis of complex combustion systems such as afterburners, rocket engines and multi-injection nozzle combustion chambers.

[0066] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0067] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A high-speed image analysis method for the interaction structure between multiple jet flames, characterized in that: The following steps are involved: S1. Image acquisition: outside the fuel injection port of the multi-injection port combustion chamber, a high-speed video of multi-jet flame self-luminescence is obtained by using a high-speed camera, and the high-speed video is processed by an image interface program to obtain a plurality of single-frame images, wherein the flame self-luminescence includes methyl self-luminescence and hydroxyl self-luminescence; S2. Analysis area selection: selecting the analysis area for the single frame image, excluding the area upstream of the fuel injection port during the selection process; S3, initial division of the analysis area: according to the spatial position of each fuel injection port, the analysis area is initially divided, and the center lines of the two injection ports are used as a reference, and the upper and lower center lines and the upstream and downstream boundaries of a certain injection port are surrounded to form an initial rectangular analysis area; S4, Gaussian filtering: using Gaussian filtering to smooth the initial rectangular analysis area to reduce image noise; S5, feature corner point detection: perform feature corner point detection on the current frame image after Gaussian filtering to obtain a special corner point set that can represent key feature points in the flame flow process; S6, feature corner point supplement: repeat the operation of step S5 to process the next frame of image, obtain a feature corner point set, and compare it with the special corner point set of the previous frame of image, identify the feature corner points that are far away from the feature corner points of the previous frame, and save these feature corner points to the corner point set of the current frame; S7, optical flow field calculation: match the feature corner points in two adjacent frames, track the displacement of each feature corner point in the continuous frames, and obtain the motion vector set of each feature point in the flame; S8, flame region identification: analyzing the acquired motion vector set, judging and classifying the flame structure according to the motion direction and speed, judging the flame region to which each characteristic corner point belongs, so that the characteristic corner points in the current field form flame point clouds of different flame sources; S9, flame outer contour recognition: performing concave hull outer contour recognition on each flame point cloud to extract the outer contour shape of the flame; S10, adjacent flame interaction extraction: perform bitwise AND operation on each flame point cloud to obtain the interaction area between adjacent flames, and finally obtain the interaction structure between multiple jet flames.

2. The high-speed image analysis method for the interaction structure between multiple jet flames according to claim 1, characterized in that: In step S5, the goodFeatureToTrack algorithm is used to perform feature corner point detection on the current frame image after Gaussian filtering.

3. The high-speed image analysis method for the interaction structure between multiple jet flames according to claim 1, characterized in that: In step S7, the Lucas-Kanade pyramid iteration algorithm is used to match the feature corner points in two adjacent image frames.

4. The high-speed image analysis method for the interaction structure between multiple jet flames according to claim 1, characterized in that: In step S8, when determining the flame area to which each characteristic corner point belongs, if it is the first determination, the determination is made according to the rectangular space where the characteristic corner point is located; If it is a subsequent judgment, the judgment is made based on the flame source to which the upstream characteristic corner point of the vector belongs.

5. The high-speed image analysis method for the interaction structure between multiple jet flames according to claim 1, characterized in that: In step S9, the Moreira-Santos algorithm based on K-means clustering is used to perform flame contour recognition.

6. An application of the high-speed image analysis method for the interaction structure between multiple jet flames as claimed in any one of claims 1 to 5, characterized in that: The high-speed image analysis method is used to extract and analyze the interaction structure between radially distributed multiple flames.

7. The use according to claim 6, characterized in that The radially distributed multiple flames are afterburner flames.

8. The use according to claim 6, characterized in that The radially distributed multiple flames are rocket engine flames.

9. The use according to claim 6, characterized in that The radially distributed multiple flames are multi-injection port combustion chamber flames.

Images