Design method for analyzing time sequence relative flow field in variable-speed pump

Abstract

The invention discloses a design method for analyzing a time sequence relative flow field in a variable-speed pump, which comprises the following steps of: 1) debugging an experimental test system for visual observation of a flow field in a centrifugal pump, and setting the sampling frequency of a high-frequency PIV (Particle Image Velocimetry) system; 2) identifying features on the impeller and constructing main features of the impeller by adopting a particle image blade identification method written by Matlab; 3) calculating phase angle differences between all adjacent time sequence particle images, and then determining phase angle differences between blade positions in all the particle images and blade positions in the first frame of particle image; 4) rotating the positions of the blades in all particle images to the position of the impeller blade in the first frame of particle image by adopting an image rotation method; 5, PIVLab particle image analysis software based on Matlab software is adopted, and all rotated particle images are computed.According to the method, knowledge and technologies in the subject fields of image processing, fluid mechanics and the like are combined, and finally high-precision calculation of the relative velocity field in the impeller of the variable-speed centrifugal pump is achieved.

Description

Technical Field

[0001] The invention relates to the fields of image processing and fluid mechanics, and in particular to a design method for analyzing a time-series relative flow field in a variable speed pump. Background Art

[0002] Centrifugal pumps have structural features such as strong rotation, large curvature and multiple walls, which lead to very complex flow fields in the impeller. Especially during the start-up and shutdown process of the centrifugal pump, the change in speed will significantly increase the complexity of the flow field, resulting in serious instability of the system. Therefore, accurately capturing the time-series flow field information in the impeller is crucial to understanding the formation mechanism of these complex flow phenomena and the optimal design of the pump station system.

[0003] PIV technology is a flow field measurement technology based on particle images. By analyzing the captured particle images, velocity information in the flow field can be obtained. This technology has the advantages of non-invasiveness and high temporal and spatial resolution, and has been widely used in fluid mechanics research. However, in the impeller of a variable speed centrifugal pump, the processing of sequential PIV particle images and the calculation of relative velocity fields still face some challenges.

[0004] First, the impeller speed changes with time during the start-up and shutdown of the centrifugal pump, and the direct result obtained based on PIV is the absolute velocity field of the flow field in the impeller. When the impeller speed is unknown, it is impossible to calculate the relative velocity field results in the impeller through the traditional velocity triangle method, thereby analyzing the morphological characteristics of the complex flow phenomena in the impeller.

[0005] Secondly, in the past experimental observation studies of flow fields in centrifugal pumps, most researchers used standard commercial PIV systems. This PIV system has relatively mature post-processing software, but its sampling frequency is only 15Hz. Since the speed of the centrifugal pump impeller is usually very high, this PIV system can only obtain an instantaneous flow field when the impeller returns to the initial impeller phase after several rotation cycles. Therefore, only the phase-locking technology can be used to obtain the phase-averaged flow field in the impeller. However, the flow field in the centrifugal pump has strong transient characteristics, and the phase averaging process will model the rich pulsating flow field characteristics in the impeller, and sometimes even obtain worthless flow field results, such as when rotational stall occurs in the impeller. Therefore, the traditional PIV system cannot effectively observe the continuous evolution of complex flow phenomena in the impeller of the centrifugal pump.

[0006] In the existing visualization observation research of flow field in centrifugal pumps, the sampling frequency of the PIV system is relatively low, while the speed of the centrifugal pump impeller is relatively high, which makes it impossible to accurately obtain the time-series flow field information in the centrifugal pump, especially for the continuous evolution of complex flow phenomena in the impeller under small flow conditions. In addition, the phase-averaged flow field processing method will model the rich transient flow field characteristics in the centrifugal pump, which is one of the main reasons why the mechanism of various complex flow phenomena in the centrifugal pump cannot be clearly revealed. With the continuous development of hardware equipment such as lasers and high-speed cameras, the research and development of high-frequency PIV systems has become possible. However, the particle image processing method in the centrifugal pump impeller under this high-frequency PIV system is not mature, especially it cannot accurately analyze the instantaneous relative velocity flow field characteristics in the variable speed impeller. Summary of the invention

[0007] The purpose of the present invention is to overcome the defects of the prior art and provide a design method for analyzing the time-series relative flow field in a variable speed pump. By combining the knowledge and technology in the fields of image processing, fluid mechanics, etc., high-precision calculation of the relative velocity field in the impeller of a variable speed centrifugal pump is finally achieved.

[0008] The object of the present invention is achieved by: a design method for analyzing the relative flow field of the timing sequence in a variable speed pump, comprising the following steps:

[0009] 1) Debug the experimental test system for visualization observation of the flow field in the centrifugal pump, set the sampling frequency of the high-frequency PIV system, and obtain the time-series particle image during the startup process of the variable speed impeller based on the entire experimental system;

[0010] 2) Using the particle image blade recognition method written in Matlab, the features on the impeller are identified and the main features of the impeller are constructed. The angles of the blade rotation between two adjacent frames of particle images are determined by identifying the features;

[0011] 3) After calculating the phase angle differences between all adjacent time-series particle images, determine the phase angle difference between the leaf position in all particle images and the leaf position in the first frame particle image;

[0012] 4) Using an image rotation method, the positions of the blades in all particle images are rotated to the positions of the impeller blades in the first frame particle image, so that the positions of the blades in all particle images coincide;

[0013] 5) The Matlab-based PIVLab particle image analysis software is used to calculate all the rotated particle images, and finally the time-series relative velocity field in the impeller of the variable speed centrifugal pump is obtained.

[0014] As a further limitation of the present invention, step 1) includes: during the start and stop process of the centrifugal pump, within a time period T, the speed of the impeller changes from N1 to N2; when the time is T(i), the speed of the impeller is N(i), firstly setting the highest PIV sampling frequency f max The time-series particle images in the impeller are collected. At this time, the interval between two instantaneous particle images is dt. Since the rotation speed N of the impeller is changing during this process, the angle of the blade rotation in two adjacent frames of particle images is changing. In a series of particle images, at time T(i), the angle interval of the blades in adjacent particle images is θ(i).

[0015] As a further limitation of the present invention, the step 2) comprises: using a particle image feature recognition method written based on Matlab software to obtain the coordinates of the feature point position in the initial particle image as (x0, y0), and then rotating the position of the feature point in the particle image to the coordinates (x1, y1), (x2, y2) ... (x i ,y i ); the phase of the blade in the i-th frame particle image and the angle interval of the blade in the i-1-th frame particle image is Δθ(i)=θ(i)-θ(i-1), where tan(θ(i))=y i / x i ,θ(i)=arctan(y i / x i );tan(θ(i-1))=y i-1 / x i-1 ,θ(i-1)=arctan(y i-1 / x i-1 );

[0016] The main calculation procedures for image feature point recognition are:

[0017] img = imread('image.jpg'); read image

[0018] Harris corner detection, display the image and corner points, and mark the coordinates:

[0019] [corners,values]=harris(img);

[0020] figure;

[0021] imshow(img);

[0022] hold on;

[0023] for i=1:size(corners,1)

[0024] plot(corners(i,1),corners(i,2),'r*');

[0025] text(corners(i,1),corners(i,2),['('num2str(corners(i,1))','num2str(corners(i,2))')']); Determine the coordinates of the feature points in the image.

[0026] As a further limitation of the present invention, step 3) comprises: calculating the blade phase angle interval △θ(i) in all adjacent particle images, and obtaining the phase difference between the blade phase in the i-th frame particle image and the blade phase in the first frame particle image.

[0027] As a further limitation of the present invention, step 4) comprises: based on the position of the initial impeller blade phase in the first frame particle image and the angular interval between the position of the blade phase in all particle images and the initial impeller phase Based on the Matlab program, the positions of the impeller blades in all particle images were rotated to the positions of the initial blades so that the blade areas in all particle images overlapped;

[0028] The key particle image rotation implementation program is:

[0029] str1 = 'L0'; str2 = num2str(n); str3 = '.bmp'; Image name expression method

[0030] SC = [str1, str2, str3]; the name of the image

[0031] i = imread(SC); read image

[0032] The rotation angle of the i-th image

[0033] figure(1)

[0034] h3 = imshow(rotateAround(imread(SC),x,y,j)); Use the rotateAround function to rotate the image, where x and y represent the coordinates of the center of the impeller.

[0035] f = getframe(figure(1));

[0036] imwrite(h3.CData,['Folder location where the rotated image is stored\',num2str(n),'.bmp']).

[0037] As a further limitation of the present invention, step 5) includes: assuming that when the highest PIV sampling frequency f is used max When observing the flow field in the impeller, the time interval between adjacent particle images is dt. A series of particle images with different interval sizes are selected to form samples for calculating the instantaneous flow field in the impeller with high precision. When the particle image interval is n, the velocity V(n) is calculated. When the velocity ΔV(n) between adjacent particle images is the smallest, the interval n is selected as the optimal particle image interval for calculating the relative velocity field:

[0038] ΔV(n)=V(n)-V(n-1).

[0039] The present invention adopts the above technical scheme, and compared with the prior art, the beneficial effects are: 1) High precision and high resolution: The present invention can achieve high-precision and high-resolution calculation of the relative velocity field in the impeller of a variable speed centrifugal pump, thereby providing more accurate flow information for the evolution characteristics of complex flow phenomena during the start-stop process of the centrifugal pump.

[0040] 2) Strong adaptability: The present invention can adapt to variable speed centrifugal pumps of different types and specifications and has strong adaptability.

[0041] 3) Saving computing resources: The present invention reduces the resources required for relative velocity field calculation by optimizing the calculation procedure of particle images, making it more suitable for practical applications.

[0042] 4) Reduced construction cost of high-frequency PIV system: The present invention can significantly reduce the minimum sampling frequency of the required PIV system and reduce the construction cost of the high-frequency PIV system. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is the experimental test system of the present invention.

[0044] Figure 2 This is a working principle diagram of the high-frequency PIV system of the present invention.

[0045] Figure 3 Schematic diagram of the process of the present invention.

[0046] Figure 4 Schematic diagram of several typical blade phase positions in the time-series PIV particle image collected in the variable speed centrifugal pump impeller of the present invention.

[0047] Figure 5 Schematic diagram of the method for identifying the impeller rotation angle between adjacent particle images under high-frequency PIV of the present invention.

[0048] Figure 6 Comparison between the original particle image and the particle image after image rotation.

[0049] Figure 7 It is a schematic diagram of the main calculation steps of the relative velocity field in the variable speed impeller in the present invention.

[0050] Figure 8 It is a schematic diagram of the dynamic initiation process of the stall vortex in the five-blade impeller of the centrifugal pump of the present invention when the impeller speed gradually increases. DETAILED DESCRIPTION

[0051] like Figure 1 As shown, the present invention first builds a single-exposure high-frequency PIV testing system, through which the time-series particle images in the impeller of a variable speed centrifugal pump are continuously collected, and then the present invention is used to calculate the relative velocity field of the time-series PIV particle images in the impeller of the variable speed centrifugal pump, and finally the evolution process of the complex flow phenomenon in the variable speed impeller is obtained.

[0052] The high-frequency PIV test system includes key components such as lasers, high-speed cameras and synchronizers. The laser is a Beijing Zhongke Siyuan DCQ-30 kHz PIV green laser with a laser wavelength of 527nm. It can work in single-pulse or double-pulse mode with a repetition frequency range of 1-20kHz. The high-speed camera is a Shenzhen Kailit nac high-speed camera. When the frame is set to 1920×1080 pixels, the sampling frequency can reach 20kHz. The synchronizer uses the ZK Model DG538 laser pulse synchronizer to ensure synchronization accuracy, coordinate and control the working sequence of the laser, high-speed camera and acquisition software, and jointly complete the image acquisition work. The working principle of the high-frequency PIV system is as follows: Figure 2 As shown in the figure, after the synchronizer is triggered, the laser emits continuous pulses with the same time interval Δt after a delay of Δt0, and the high-speed camera synchronously collects continuous particle images. Based on two consecutive frames of particle images, an instantaneous flow field can be obtained. The time interval between two adjacent frames of particle images is dt, and at this time dt = Δt.

[0053] like Figure 3 A design method for analyzing the relative flow field of the timing sequence in a variable speed pump is shown, characterized in that it includes the following steps:

[0054] 1) Debug the experimental test system for visualization observation of the flow field in the centrifugal pump, set the sampling frequency of the high-frequency PIV system, and obtain the time-series particle image during the startup process of the variable speed impeller based on the entire experimental system;

[0055] During the start-up and shutdown process of the centrifugal pump, the impeller speed changes from N1 to N2 within a time period T; when the time is T(i), the impeller speed is N(i), and the highest PIV sampling frequency f is first set. maxCollect the time-series particle images in the impeller. At this time, the interval between two instantaneous particle images is dt. Since the speed N of the impeller changes during this process, the angle of the blades in two adjacent frames of particle images changes. In a series of particle images, at time T(i), the angle interval of the blades in adjacent particle images is θ(i). Examples of the phase positions of the impeller blades at several typical different times are shown below. Figure 4 shown.

[0056] 2) The particle image blade recognition method written in Matlab is used to identify the features on the impeller and construct the main features of the impeller. The angle of rotation of the blades in two adjacent frames of particle images is determined by identifying the features, such as Figure 5 As shown;

[0057] The particle image feature recognition method based on Matlab software is used to obtain the coordinates of the feature point position in the initial particle image as (x0, y0), and then the position of the feature point in the particle image is rotated to the coordinates (x1, y1), (x2, y2)…(x i ,y i ); the phase of the blade in the i-th frame particle image and the angle interval of the blade in the i-1-th frame particle image is Δθ(i)=θ(i)-θ(i-1), where tan(θ(i))=y i / x i ,θ(i)=arctan(y i / x i );tan(θ(i-1))=y i-1 / x i-1 ,θ(i-1)=arctan(y i-1 / x i-1 );

[0058] The main calculation procedures for image feature point recognition are:

[0059] img = imread('image.jpg'); read image

[0060] Harris corner detection, display the image and corner points, and mark the coordinates:

[0061] [corners,values]=harris(img);

[0062] figure;

[0063] imshow(img);

[0064] hold on;

[0065] for i=1:size(corners,1)

[0066] plot(corners(i,1),corners(i,2),'r*');

[0067] text(corners(i,1),corners(i,2),['('num2str(corners(i,1))','num2str(corners(i,2))')']); Determine the coordinates of the feature points in the image.

[0068] 3) After calculating the phase angle differences between all adjacent time-series particle images, determine the phase angle difference between the leaf position in all particle images and the leaf position in the first frame particle image;

[0069] By calculating the blade phase angle interval △θ(i) in all adjacent particle images, the phase difference between the blade phase in the i-th frame particle image and the blade phase in the first frame particle image is obtained.

[0070] 4) Use the image rotation method to rotate the positions of the blades in all particle images to the positions of the impeller blades in the first frame particle image, so that the positions of the blades in all particle images coincide. Figure 6 (a) shows the original particle image pair. It can be found that there is a blade area in the query window. The existence of the blade area will cause deviations in the flow field calculation results. Figure 6 (b) shows the rotated particle image pair. It can be found that the query window in the rotated particle image does not contain the blade area, which can ensure that the flow field calculation results near the blade are more accurate.

[0071] Based on the position of the initial impeller blade phase in the first frame particle image and the angular interval between the position of the blade phase in all particle images and the initial impeller phase Based on the Matlab program, the positions of the impeller blades in all particle images were rotated to the positions of the initial blades so that the blade areas in all particle images overlapped;

[0072] The key particle image rotation implementation program is:

[0073] str1 = 'L0'; str2 = num2str(n); str3 = '.bmp'; Image name expression method

[0074] SC = [str1, str2, str3]; the name of the image

[0075] i = imread(SC); read image

[0076] The rotation angle of the i-th image

[0077] figure(1)

[0078] h3 = imshow(rotateAround(imread(SC),x,y,j)); Use the rotateAround function to rotate the image, where x and y represent the coordinates of the center of the impeller.

[0079] f = getframe(figure(1));

[0080] imwrite(h3.CData,['Folder location where the rotated image is stored\',num2str(n),'.bmp']).

[0081] 5) PIVLab particle image analysis software based on Matlab software is used to calculate all the rotated particle images, and finally the time series relative velocity field in the impeller of the variable speed centrifugal pump is obtained as follows: Figure 7 As shown, Figure 7 (a) is the position of the blade in two adjacent frames of particle images obtained by the high-frequency PIV system. The two frames of particle images are made to overlap by image rotation. Figure 7 (b) Then, the PIVlab software is used to calculate the image of the rotated particles, and the relative velocity field in the impeller can be obtained. Figure 7 (c).

[0082] Assume that when the highest PIV sampling frequency f is used max When observing the flow field in the impeller, the time interval between adjacent particle images is dt. For example, when the interval number 1 is selected to extract the particle image, the time interval between two adjacent frames of particle images is 2dt. A series of particle images with different interval sizes are selected to form samples for calculating the instantaneous flow field in the impeller with high precision. When the particle image interval is n, the velocity V(n) is calculated at this time. When the velocity ΔV(n) between adjacent particle images is the smallest, the interval n is selected as the optimal particle image interval for calculating the relative velocity field:

[0083] ΔV(n)=V(n)-V(n-1).

[0084] Based on the above method, the time-series transient velocity results in the variable speed impeller under the high-frequency PIV technology are obtained. This method can obtain the relative velocity field results in the impeller when the impeller speed is unknown.

[0085] like Figure 8 As shown in Figure 1, the dynamic initiation process of the stall vortex in the five-blade impeller during the gradual increase of the impeller speed of the centrifugal pump is shown in Figure 2, where t is the sampling time and T is the impeller period. Figure 8From (a) to (c), it can be found that the small-sized high-frequency shed vortex at the inlet of flow channel B moves along the suction side to the middle of the blade, and its shape gradually decreases and finally disappears, as shown in Figure 8 (d) to (e). Figure 8 From (f) to (i), the flow field on the suction side of the blade in channel B undergoes a similar vortex evolution process. Figure 8 As shown in (i), the flow field in the impeller channel is relatively stable. However, after one cycle, Figure 8 As shown in (j), a large vortex suddenly appears near the outlet of the blade suction side, and a small vortex also exists at the inlet of the flow channel. There is still a short smoothing area between the two vortices on the blade suction side. In the next two impeller cycles, the two vortices begin to merge with each other. Figure 8 (k) to (l). The present invention clearly demonstrates the dynamic formation process of the primary stall vortex in the impeller flow channel of the centrifugal pump impeller during the start-up and shutdown process. Therefore, the present invention can provide a key technical means for revealing the formation mechanism of complex flow phenomena in the centrifugal pump.

[0086] The present invention aims at the characteristics of the particle image in the impeller, and uses Matlab software to construct a set of image rotation methods. First, it is necessary to determine the rotation speed of the impeller and the center of the impeller in the particle image, and then all the particle images are rotated to the same impeller phase through the image rotation method. The rotated particle image is directly calculated by PIVlab software, and the result after calculation is the relative velocity field in the impeller. This method can not only improve the solution accuracy of the flow field near the blade, but also reduce the minimum sampling frequency required by the high-frequency PIV system by removing the displacement of the blade in the circumferential direction in the particle image, thereby reducing the construction cost of the high-frequency PIV system.

[0087] However, the unknown speed of the variable speed impeller will result in an unknown phase difference between two adjacent frames of images within the time interval of Δt, so the relative velocity field cannot be obtained by the image rotation method. To solve this problem, the present invention will further construct a particle image recognition and tracking method, which first determines the phase difference between two adjacent frames of the variable speed impeller image through the tracking method, and then uses the image rotation method to make the blade areas in the two frames of particle images overlap, and finally the relative velocity field in the impeller can be obtained.

[0088] The present invention is not limited to the above-mentioned embodiments. On the basis of the technical solution disclosed in the present invention, technicians in this field can make some substitutions and deformations to some technical features therein according to the disclosed technical content without creative labor, and these substitutions and deformations are all within the protection scope of the present invention.

Claims

1. A design method for analyzing the relative flow field of the timing sequence in a variable speed pump, characterized in that: The following steps are involved: 1) Debug the experimental test system for visualization observation of the flow field in the centrifugal pump, set the sampling frequency of the high-frequency PIV system, and obtain the time-series particle image during the startup process of the variable speed impeller based on the entire experimental system; 2) Using the particle image blade recognition method written in Matlab, the features on the impeller are identified and the main features of the impeller are constructed. The angles of the blade rotation between two adjacent frames of particle images are determined by identifying the features; 3) After calculating the phase angle differences between all adjacent time-series particle images, determine the phase angle difference between the leaf position in all particle images and the leaf position in the first frame particle image; 4) Using an image rotation method, the positions of the blades in all particle images are rotated to the positions of the impeller blades in the first frame particle image, so that the positions of the blades in all particle images coincide; 5) The Matlab-based PIVLab particle image analysis software is used to calculate all the rotated particle images, and finally the time-series relative velocity field in the impeller of the variable speed centrifugal pump is obtained.

2. A design method for analyzing the time-series relative flow field in a variable speed pump according to claim 1, characterized in that: The step 1) includes: during the start and stop process of the centrifugal pump, within a time period T, the speed of the impeller changes from N1 to N2; when the time is T(i), the speed of the impeller is N(i), firstly setting the highest PIV sampling frequency f max The time-series particle images in the impeller are collected. At this time, the interval between two instantaneous particle images is dt. Since the rotation speed N of the impeller is changing during this process, the angle of the blade rotation in two adjacent frames of particle images is changing. In a series of particle images, at time T(i), the angle interval of the blades in adjacent particle images is θ(i).

3. A design method for analyzing the timing relative flow field in a variable speed pump according to claim 1, characterized in that: The step 2) includes: using a particle image feature recognition method written based on Matlab software to obtain the coordinates of the feature point position in the initial particle image as (x0, y0), and then rotating the position of the feature point in the particle image to the coordinates (x1, y1), (x2, y2) ... (x i ,y i ); the phase of the blade in the i-th frame particle image and the angle interval of the blade in the i-1-th frame particle image is Δθ(i)=θ(i)-θ(i-1), where tan(θ(i))=y i / x i ,θ(i)=arctan(y i / x i );tan(θ(i-1))=y i-1 / x i-1 ,θ(i-1)=arctan(y i-1 / x i-1 ); The main calculation procedures for image feature point recognition are:

4. A design method for analyzing the time-series relative flow field in a variable speed pump according to claim 1, characterized in that: The step 3) comprises: calculating the blade phase angle interval △θ(i) in all adjacent particle images, and obtaining the phase difference between the blade phase in the i-th frame particle image and the blade phase in the first frame particle image.

5. A design method for analyzing the relative flow field of the timing sequence in a variable speed pump according to claim 1, characterized in that: The step 4) comprises: based on the position of the initial impeller blade phase in the first frame particle image and the angular interval between the position of the blade phase in all particle images and the initial impeller phase Based on the Matlab program, the positions of the impeller blades in all particle images were rotated to the positions of the initial blades so that the blade areas in all particle images overlapped; The key particle image rotation implementation program is: str1 = 'L0'; str2 = num2str(n); str3 = '.bmp'; Image name expression method SC = [str1, str2, str3]; the name of the image i = imread(SC); read image The rotation angle of the i-th image figure(1) h3 = imshow(rotateAround(imread(SC),x,y,j)); Use the rotateAround function to rotate the image, where x and y represent the coordinates of the center of the impeller. f = getframe(figure(1)); imwrite(h3.CData,['Folder location where the rotated image is stored\',num2str(n),'.bmp']).

6. A design method for analyzing the relative flow field of the timing sequence in a variable speed pump according to claim 1, characterized in that: The step 5) includes: assuming that when the highest PIV sampling frequency f is adopted max When observing the flow field in the impeller, the time interval between adjacent particle images is dt. A series of particle images with different interval sizes are selected to form samples for calculating the instantaneous flow field in the impeller with high precision. When the particle image interval is n, the velocity V(n) is calculated. When the velocity ΔV(n) between adjacent particle images is the smallest, the interval n is selected as the optimal particle image interval for calculating the relative velocity field: ΔV(n)=V(n)-V(n-1).

Images