Machine Vision-Based Dynamic Balance Evaluation Method for Axial Flow Fan Impellers

Abstract

This invention relates to the field of image processing technology and proposes a machine vision-based method for evaluating the dynamic balance of axial flow fan impellers. The method includes: acquiring continuous frames of impeller images of a high-speed rotating axial flow fan; obtaining impeller edge images for each frame corresponding to the analysis period and several moments within that period; extracting several key points from the impeller edge images; obtaining the impeller load non-uniformity during the analysis period; obtaining the dynamic instability index during the analysis period; obtaining the impeller dynamic balance deviation during the analysis period; obtaining the impeller axial deflection index during the analysis period; obtaining the impeller structural anomaly index during the analysis period; obtaining the visual fusion dynamic balance index of the impeller during the analysis period; obtaining the adjusted dynamic balance threshold during the analysis period; and performing a dynamic balance evaluation of the impeller. This invention aims to solve the problem that high-speed impeller rotation leads to image blurring, making accurate dynamic balance evaluation through trajectory analysis difficult.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and more specifically to a machine vision-based method for evaluating the dynamic balance of axial flow fan impellers. Background Technology

[0002] Axial flow fans are widely used in industrial settings such as mines, tunnels, subways, power plants, and petrochemical plants. They are critical gas conveying and ventilation equipment. The impeller is the core component of the axial flow fan, and its dynamic balance directly affects the overall vibration level, service life, energy efficiency rating, and operational safety. An impeller with poor dynamic balance may cause equipment damage or foundation loosening, and long-term eccentric operation can lead to premature bearing fatigue, shortening the fan's lifespan. Therefore, efficient, accurate, and real-time dynamic balance assessment of axial flow fan impellers is an important means to ensure the reliability of ventilation systems, extend equipment lifespan, and reduce maintenance costs. In recent years, with the development of machine vision technology, especially the maturity of high-speed image acquisition, image registration, 3D reconstruction, and intelligent image analysis algorithms, non-contact dynamic balance assessment methods based on machine vision have gradually become a research hotspot.

[0003] Machine vision has advantages in dynamic balancing inspection because it is a non-contact measurement method that does not affect the normal operation of the fan. Image recognition can achieve high-precision capture of rotational state and detection of minute eccentricities, and can be deployed on the production line to achieve continuous monitoring and closed-loop control. However, it is difficult to reliably extract the trajectory information of key feature points during impeller rotation in high-speed rotation environments. Due to the high-speed rotation of the impeller, problems such as image blurring, feature drift, and trajectory breakage can occur. Traditional image acquisition and trajectory analysis algorithms cannot obtain a representative rotation path within a limited time window, thus affecting the accuracy of dynamic balancing assessment. In addition, on-site factors such as uneven lighting, background interference, and feature occlusion can also interfere with the interpretation of visual images, causing misjudgments or omissions. Summary of the Invention

[0004] This invention provides a machine vision-based method for dynamic balance assessment of axial flow fan impellers, addressing the problem that existing methods for accurate dynamic balance assessment through trajectory analysis result in blurred images due to high-speed impeller rotation. The specific technical solution adopted is as follows:

[0005] This invention proposes a machine vision-based method for evaluating the dynamic balance of axial flow fan impellers, which includes the following steps:

[0006] Acquire continuous frames of impeller images of a high-speed rotating axial flow fan, obtain impeller edge images for each frame corresponding to the analysis period and several moments within it, extract several key points from the impeller edge images, and record parameter data of several key operating conditions of the axial flow fan at each moment and the temperature at several locations in the impeller.

[0007] Based on the trajectory smoothing of key points in the impeller edge images of consecutive frames during the analysis period, and the area distribution of the closed regions formed by the trajectories, the impeller load non-uniformity during the analysis period is obtained. Combining the changes in the area difference of different blades in the impeller edge images of consecutive frames, the dynamic instability index during the analysis period is obtained. Based on the changes in the dynamic instability index between adjacent time periods, and the differences in the temperature difference change trends at different positions of the impeller, combined with the changes in the parameter data fluctuations of each key operating condition, the impeller dynamic balance deviation during the analysis period is obtained.

[0008] Based on the changes in the position of the centroid and the diameter of the trajectory region formed by the trajector edge images of key points in consecutive frames during the analysis period, the axial deflection index of the impeller during the analysis period is obtained; combined with the changes in the grayscale performance of each blade region in adjacent frames of impeller images, the impeller structural anomaly index of the analysis period is obtained; based on the impeller dynamic balance deviation and the impeller structural anomaly index during the analysis period, the visual fusion dynamic balance index of the impeller during the analysis period is obtained.

[0009] The initial dynamic balance threshold of the impeller is adjusted based on the visual fusion dynamic balance index, the adjusted dynamic balance threshold for the analysis period is obtained, and the dynamic balance of the impeller is evaluated.

[0010] Optionally, the specific method for obtaining the impeller load non-uniformity during the analysis period includes:

[0011] The position of any key point in each frame of the impeller edge image during the analysis period is fitted to obtain the motion trajectory of the key point during the analysis period, and the closed area formed by the closed motion trajectory is obtained; the closed area is divided into a left area and a right area by a perpendicular line through the impeller center point, and the absolute value of the area difference between the left area and the right area is obtained as the closed area difference of the key point during the analysis period.

[0012] The direction from the previous position to the next position in any adjacent frame of the impeller edge image during the analysis period is obtained as the motion direction of the key point in the previous frame of the adjacent frame during the analysis period; the motion direction of the key point in each frame during the analysis period is obtained, and the product of the standard deviation of the angle between the motion directions of the key point in all adjacent frames during the analysis period and the difference in the closed area is used as the trajectory non-uniformity of the key point during the analysis period.

[0013] The mean of the trajectory non-uniformity of all key points during the analysis period is taken as the impeller load non-uniformity during the analysis period.

[0014] Optionally, the specific method for obtaining the dynamic instability index for the analysis period includes:

[0015] For any frame of impeller edge image in the analysis period, obtain several blade regions in the frame of impeller edge image and obtain the area of ​​each blade region. Obtain the absolute value of the difference between the areas of any two blade regions. Use the mean of the absolute values ​​of the differences between the areas of all pairs of blade regions in the frame of impeller edge image as the blade area difference factor of the frame of impeller edge image.

[0016] Obtain the absolute value of the difference in blade area difference factor between any two adjacent frames of impeller edge images during the analysis period. Multiply the mean of the absolute values ​​of the difference in blade area difference factor between all adjacent frames of impeller edge images during the analysis period by the impeller load non-uniformity during the analysis period, and use this product as the dynamic instability index for the analysis period.

[0017] Optionally, the specific method for obtaining the impeller dynamic balance deviation during the analysis period includes:

[0018] Based on the changes in the dynamic instability index between adjacent time periods and the differences in the temperature difference trends at different locations of the impeller, the impeller load trend index for the analysis period is obtained.

[0019] Obtain the standard deviation of all parameter data for any key operating condition during the analysis period, obtain the standard deviation of all parameter data for the same key operating condition in the adjacent previous period of the analysis period, and use the ratio of the standard deviation of the key operating condition in the analysis period to the standard deviation of the same key operating condition in the adjacent previous period as the parameter fluctuation change factor for the key operating condition during the analysis period.

[0020] Based on the impeller load trend index during the analysis period and the mean of the parameter fluctuation change factors for each key operating condition during the analysis period, the impeller dynamic balance deviation during the analysis period is obtained. The impeller dynamic balance deviation is positively correlated with the impeller load trend index and negatively correlated with the mean of the parameter fluctuation change factors for each key operating condition.

[0021] Optionally, the specific method for obtaining the impeller load trend index for the analysis period includes:

[0022] For the temperature at several locations in the impeller at any given moment during the analysis period, the absolute value of the temperature difference between any two locations is obtained. The mean of the absolute values ​​of the temperature differences between all pairs of locations in the impeller at that moment is taken as the mean temperature difference at that moment. A coordinate system is constructed with time as the horizontal axis and temperature difference as the vertical axis. The mean temperature difference at each moment during the analysis period is mapped onto the coordinate system to obtain several data points. The data points are then fitted using the least squares method to obtain a fitted straight line and its slope. The slope is taken as the temperature difference variation factor during the analysis period.

[0023] The ratio of the temperature difference change factor in the analysis period to the temperature difference change factor in the adjacent previous period is obtained, and the ratio of the dynamic instability index in the analysis period to the dynamic instability index in the adjacent previous period is obtained. The product of the two ratios is used as the impeller load trend index in the analysis period.

[0024] Optionally, the impeller axial deflection index during the analysis period is obtained using the following method:

[0025] For any key point in the analysis period, the position of the impeller edge image in each frame is connected to the position in the impeller edge image of the adjacent frame to obtain the movement trajectory of the key point in the analysis period. The movement trajectory forms several closed regions, which are denoted as several trajectory regions. The centroid of each trajectory region and the length of the horizontal straight line passing through the centroid in the trajectory region are obtained as the horizontal diameter of each trajectory region.

[0026] Based on the centroid distribution and horizontal diameter changes in the trajectory region of key points, the axial deflection coefficient of each key point during the analysis period is obtained.

[0027] The average of the axial deflection coefficients of all key points during the analysis period is taken as the impeller axial deflection index for the analysis period.

[0028] Optionally, the axial deflection coefficients of each key point during the analysis period are obtained using the following method:

[0029] For any key point in the analysis period, obtain the distance between the centroids of any two trajectory regions. Arrange the horizontal diameters of all trajectory regions of the key point in ascending order to obtain the horizontal diameter sequence of the key point. Obtain the ratio of the latter horizontal diameter to the former horizontal diameter in the horizontal diameter sequence. Multiply the mean of the ratios of all adjacent horizontal diameters in the horizontal diameter sequence with the standard deviation of the distance between the centroids of all pairwise trajectory regions of the key point as the axial deflection coefficient of the key point in the analysis period.

[0030] Optionally, the specific method for obtaining the impeller structure anomaly index for the analysis period includes:

[0031] For any frame of impeller image in the analysis period, obtain the number of blade regions in the frame of impeller image, and the average of the average gray values ​​of pixels in each blade region. The product of the number of blade regions and the average of the average gray values ​​is used as the gray visual performance value of the frame of impeller image. Obtain the absolute value of the difference between the gray visual performance values ​​of any adjacent frame of impeller image in the analysis period.

[0032] Based on the impeller axial deflection index during the analysis period and the mean of the absolute values ​​of the differences in grayscale visual representation values ​​of all adjacent frames of the impeller image during the analysis period, the impeller structure anomaly index for the analysis period is obtained. The impeller structure anomaly index is positively correlated with the impeller axial deflection index and negatively correlated with the mean of the absolute values ​​of the differences in grayscale visual representation values.

[0033] Optionally, the specific method for obtaining the visual fusion dynamic balance index of the impeller during the analysis period includes:

[0034] The sum of the inversely proportional normalized result of the impeller dynamic balance deviation during the analysis period and the inversely proportional normalized result of the impeller structural anomaly index during the analysis period is used as the visual fusion dynamic balance index of the impeller during the analysis period.

[0035] Optionally, the specific method for obtaining the adjustment dynamic balance threshold for the analysis period includes:

[0036] The product of the initial dynamic balance threshold of the impeller and the visual fusion dynamic balance index of the impeller during the analysis period is used as the adjustment dynamic balance threshold for the analysis period.

[0037] The beneficial effects of this invention are as follows: This invention acquires continuous frames of impeller images using a high-speed camera, analyzes the trajectory symmetry and edge smoothness of key points during impeller rotation to quantify impeller load unevenness, and combines this with the area change of the blade region to obtain a dynamic instability index, reflecting the dynamic disturbance amplitude caused by uneven load during the rotation of the axial flow fan impeller; by analyzing the trend of impeller surface temperature difference changes and the changes in the dynamic instability index in adjacent time periods, an impeller load trend index is obtained to quantify the impact of temperature changes on the material and the resulting decrease in dynamic stability. Furthermore, by combining the fluctuation changes of key operating condition parameters, the influence of other operating condition parameters is eliminated, clarifying the impeller dynamic balance deviation; then, by analyzing the axial offset of key point trajectories during rotation and the sampling error of machine vision, the structural anomaly index of the fan impeller under rotation is explained, and a visual fusion dynamic balance index is obtained by combining dynamic balance deviation. By combining two-dimensional features and inferred three-dimensional features, the initial dynamic balance threshold is adjusted, realizing the dynamic adjustment mechanism of the axial flow fan impeller dynamic balance assessment mechanism. This effectively helps relevant personnel make more accurate decisions after dynamic balance assessment, ensuring the safety and stability of equipment under various conditions. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a schematic diagram of a machine vision-based dynamic balance evaluation method for axial flow fan impellers provided in one embodiment of the present invention.

[0040] Figure 2 This is a schematic diagram of the impeller edge of an axial flow fan. Detailed Implementation

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0042] Please see Figure 1 The diagram illustrates a flowchart of a machine vision-based dynamic balance evaluation method for axial flow fan impellers, provided by an embodiment of the present invention. The method includes the following steps:

[0043] Step S001: Acquire continuous frames of impeller images of a high-speed rotating axial flow fan, obtain the impeller edge images of each frame corresponding to the analysis period and several moments within it, extract several key points from the impeller edge images, and record the parameter data of several key operating conditions of the axial flow fan at each moment and the temperature at several locations in the impeller.

[0044] The purpose of this embodiment is to acquire video stream images of the impeller rotation process of an axial flow fan using a high-speed camera, and to achieve the dynamic balance state of the impeller through visual fusion over a period of time by analyzing the dynamic balance and axial deflection of the impeller. At the same time, the influence of temperature changes of the fan impeller and parameter changes of several key operating conditions of the fan on the impeller rotation process is considered, so as to dynamically adjust the dynamic balance threshold to achieve dynamic balance assessment of the impeller rotation process.

[0045] Specifically, a high-speed industrial camera is installed at the viewing window or inspection port of the axial flow fan impeller casing. The camera should have low exposure delay and shutter synchronization capabilities to ensure that it can capture clear images of the impeller rotating at high speed. The impeller is captured as a video stream at a sampling frequency of 10 frames per second. The captured images are preprocessed to obtain consecutive frames of impeller images. The Canny edge detection algorithm is then used on each frame of the impeller image to obtain the impeller edge image, such as... Figure 2 As shown; a preset time window is used, which is described in this embodiment as 5 minutes. The time window divides several moments of the acquired impeller image into several time periods, and the most recent 5 minutes are used as the analysis period for subsequent processing; at the same time, an encoder or photoelectric trigger is installed at the end of the fan shaft or the output shaft of the motor to synchronously acquire the starting angle of each rotation, which is beneficial for the angle alignment of subsequent image frames and to realize periodic state analysis. All acquired images are temporarily stored in a buffer for synchronous calibration. The pulse signal of the photoelectric encoder or Hall sensor is connected to the acquisition system to establish the mapping relationship between image frame and rotation angle time, thereby determining the mapping relationship between impeller image and time.

[0046] Furthermore, for any frame of impeller edge image, several key points are marked at the blade tip, blade root, and central axis positions respectively. In this embodiment, a key point is marked at the blade tip, blade root, and central axis positions corresponding to each blade. In other frames of impeller edge images, key points are marked at corresponding positions according to the position changes of key points in consecutive frame images, thus obtaining several key points and their positions in consecutive frame impeller edge images.

[0047] Furthermore, a high-speed infrared camera is deployed to acquire infrared images of the fan impeller, with a sampling frequency set to once per second. Based on the infrared images, the temperature of the impeller at the blade tip, blade root, and central axis position at various times is acquired. At the same time, parameter data for several key operating conditions, including rotational speed and flow rate, are acquired at the same sampling frequency. This yields parameter data for several key operating conditions of the axial flow fan at various times and the temperature at several positions in the impeller.

[0048] It should be noted that analyzing the spatial path changes of the impeller edge and time transformation can detect signs of dynamic imbalance such as shaking, eccentricity, and oscillation. The spatial path of the edge point has high temporal resolution and sub-pixel accuracy, making it suitable for detecting minute deviations. Compared with traditional methods, it has a certain degree of reliability in analyzing the overall image.

[0049] Step S002: Based on the trajectory smoothing performance of key points in the impeller edge images of consecutive frames during the analysis period, and the area distribution of the closed region formed by the trajectory, the impeller load non-uniformity during the analysis period is obtained; combined with the changes in the area difference of different blades in the impeller edge images of consecutive frames, the dynamic instability index during the analysis period is obtained; based on the changes in the dynamic instability index of adjacent time periods, and the differences in the temperature difference change trends at different positions of the impeller, combined with the changes in the parameter data fluctuations of each key operating condition, the impeller dynamic balance deviation during the analysis period is obtained.

[0050] It should be noted that when the fan impeller is rotating, the movement trajectory of the key point is a closed area, and under normal circumstances, this closed area should be symmetrical and almost circular. By analyzing the left and right symmetry of the closed area, the uneven load on the impeller can be reflected. At the same time, the change in the direction of the trajectory reflects whether the overall movement trajectory is smooth. The more severe the fluctuation, the more irregular the movement trajectory is and the less it is close to a circle, which can also reflect the uneven load on the impeller.

[0051] Preferably, in one embodiment of the present invention, the impeller load non-uniformity during the analysis period is obtained based on the trajectory smoothing performance of key points in the impeller edge images of consecutive frames during the analysis period, and the area distribution of the closed region formed by the trajectory. The specific method includes:

[0052] The position of any key point in each frame of the impeller edge image during the analysis period is fitted to obtain the motion trajectory of the key point during the analysis period, and the closed region formed by the closed motion trajectory is obtained. The closed region is divided into a left region and a right region by a perpendicular line through the impeller center point. The absolute value of the area difference between the left region and the right region is obtained as the closed area difference of the key point during the analysis period. The direction from the previous position to the next position in any adjacent frame of the impeller edge image during the analysis period is obtained as the motion direction of the key point in the previous frame of the adjacent frame during the analysis period. The motion direction of the key point in each frame during the analysis period is obtained (the motion direction is not obtained in the last frame of the analysis period). The standard deviation of the angle between the motion directions of the key point in all adjacent frames during the analysis period is multiplied by the closed area difference as the trajectory non-uniformity of the key point during the analysis period. The mean of the trajectory non-uniformity of all key points during the analysis period is used as the impeller load non-uniformity during the analysis period.

[0053] It should be noted that a larger difference in the closed area indicates a greater difference in the area between the left and right sides, and a less symmetrical closed area, reflecting a more uneven load on the impeller. At the same time, the standard deviation of the angle between adjacent motion directions reflects the variation and fluctuation of the motion direction. A larger standard deviation indicates that the motion trajectory does not conform to the expected smooth circular area characteristics of the key points. Furthermore, by averaging all key points, a larger mean indicates that the edge points have formed asymmetrical trajectories in the movement trajectory, which means that there is a serious eccentricity or uneven load on the impeller during rotation, and the greater the unevenness of the impeller load.

[0054] It should be further noted that the load unevenness of the impeller during the analysis period refers to the possible shift of the impeller's center of force during rotation, such as mechanical eccentricity caused by uneven materials, installation deviations, local dust accumulation, etc. It does not reveal the problems generated in the impeller structure and cannot well assess the dynamic balance state of the impeller during operation.

[0055] Preferably, in one embodiment of the present invention, the dynamic instability index for the analysis period is obtained by combining the changes in the area difference of different blades in the impeller edge image across consecutive frames of impeller edge images. The specific method includes:

[0056] For any frame of the impeller edge image during the analysis period, several blade regions in that frame of the impeller edge image are obtained (directly obtained through edge detection and contour extraction), and the area of ​​each blade region is obtained. The absolute value of the difference between the areas of any two blade regions is obtained, and the mean of the absolute values ​​of the differences between the areas of all pairs of blade regions in that frame of the impeller edge image is used as the blade area difference factor of that frame of the impeller edge image. The absolute value of the difference between the blade area difference factors of any adjacent frame of the impeller edge image during the analysis period is obtained, and the mean of the absolute values ​​of the differences between the blade area difference factors of all adjacent frame of the impeller edge image during the analysis period is multiplied by the impeller load non-uniformity during the analysis period as the dynamic instability index of the analysis period.

[0057] It should be noted that the blade area hardly changes under ideal conditions during rotation. Therefore, the greater the change in the blade area difference factor, the more likely there is a deviation in the impeller during rotation, leading to large fluctuations in the blade area. Combined with the determined impeller load unevenness, this comprehensively reflects the dynamic disturbance amplitude caused by the load unevenness during the rotation of the axial flow fan impeller. It is an important composite characteristic for measuring the dynamic balance state of the impeller, that is, the greater the dynamic instability index of the fan impeller during rotation.

[0058] It should be further noted that the dynamic stability index mainly reflects the overall dynamic fluctuation trend of the impeller under the influence of various disturbances during operation. However, this macroscopic performance cannot reveal the specific reasons for the instability. Further analysis of the inertia distribution and mass eccentricity during impeller rotation can effectively identify imbalance problems caused by eccentric installation, unilateral heavy load, and bearing wear, providing valuable diagnostic basis for subsequent strategies.

[0059] Preferably, in one embodiment of the present invention, based on the changes in the dynamic instability index between adjacent time periods and the differences in the temperature difference trends at different positions of the impeller, combined with the changes in the parameter data fluctuations of each key operating condition, the impeller dynamic balance deviation for the analysis period is obtained, including the following specific methods:

[0060] For the temperature at several locations in the impeller at any given moment during the analysis period, the absolute value of the temperature difference between any two locations is obtained. The mean of the absolute values ​​of the temperature differences between all pairs of locations in the impeller at that moment is taken as the mean temperature difference at that moment. A coordinate system is constructed with time as the horizontal axis and temperature difference as the vertical axis. The mean temperature difference at each moment during the analysis period is mapped onto the coordinate system to obtain several data points. The data points are fitted using the least squares method to obtain a fitted straight line and its slope. The slope is taken as the temperature difference variation factor during the analysis period. The ratio of the temperature difference variation factor during the analysis period to the temperature difference variation factor of the adjacent previous period is obtained. The ratio of the dynamic instability index during the analysis period to the dynamic instability index of the adjacent previous period is also obtained. The product of the two ratios is taken as the impeller load trend index during the analysis period.

[0061] It should be noted that the larger the ratio of the dynamic instability index, the greater the change in the dynamic instability index during the analysis period. At the same time, the larger the ratio of the temperature difference change factor, the greater the temperature change, which may lead to changes in the physical properties of the material, such as elastic modulus, strength and fatigue life. When the temperature rises, some materials may exhibit softening, which will lead to a decrease in dynamic stability.

[0062] It should be further noted that poor dynamic balance is usually accompanied by uneven distribution of load on each blade, resulting in some blades bearing excessive force during operation. If the actual operating conditions change significantly, it may have a certain impact on the originally stable impeller, thus affecting the dynamic balance of the impeller. If the operating conditions do not change significantly, it means that the poor dynamic balance of the fan impeller may be due to some faults, and timely adjustment is required.

[0063] Furthermore, the standard deviation of all parameter data for any key operating condition during the analysis period is obtained, as well as the standard deviation of all parameter data for the same key operating condition in the adjacent previous period. The ratio of the standard deviation of the key operating condition during the analysis period to the standard deviation of the same key operating condition in the adjacent previous period is used as the parameter fluctuation factor for the key operating condition during the analysis period.

[0064] Furthermore, based on the impeller load trend index during the analysis period and the mean of the parameter fluctuation change factors for each key operating condition during the analysis period, the impeller dynamic balance deviation during the analysis period is obtained. The impeller dynamic balance deviation is positively correlated with the impeller load trend index and negatively correlated with the mean of the parameter fluctuation change factors for each key operating condition.

[0065] As an example, the ratio of the impeller load trend index during the analysis period to the mean of the parameter fluctuation change factor for each key operating condition during the analysis period is used as the impeller dynamic balance deviation during the analysis period.

[0066] It should be noted that the weights are constructed using the average of the parameter fluctuation factors of the key operating conditions. The smaller the parameter fluctuation, the less obvious the parameter data changes in the operating conditions. A higher impeller load trend index is caused by fluctuations in the dynamic balance state, that is, a sudden deterioration in the dynamic balance state, indicating that there may be a problem with the dynamic balance state of the equipment. This situation may be caused by a fault. If some parts of the impeller are worn or damaged, it may lead to uneven load distribution, thereby affecting the dynamic balance, and the greater the impeller dynamic balance deviation.

[0067] It should be further noted that after evaluating the dynamic balance deviation of the impeller during the operation of the ventilation fan in the analysis period, in the actual operating environment, this dynamic imbalance often not only causes radial vibration, but also leads to the slight tilting of the rotating shaft or axial drift in the main shaft direction. The tilting or eccentric state will cause irregular offset of the image observation point, resulting in the breakage and drift of the feature trajectory in the vertical / depth direction between image frames, further aggravating the difficulty of extracting key feature points.

[0068] Step S003: Based on the position change of the centroid and the diameter change of the trajectory region formed by the trajectory of key points in the impeller edge images of consecutive frames during the analysis period, obtain the impeller axial deflection index during the analysis period; combine the grayscale performance changes of each blade region in adjacent frame impeller images to obtain the impeller structural anomaly index during the analysis period; based on the impeller dynamic balance deviation and impeller structural anomaly index during the analysis period, obtain the visual fusion dynamic balance index of the impeller during the analysis period.

[0069] It should be noted that when analyzing the tilt and axial offset of the impeller of a ventilator in a rotating state, the dynamic balance deviation is mainly manifested as the asymmetry of the rotating mass distribution. One of the direct spatial manifestations of this asymmetry is axial tilt or central axis offset. Existing industrial cameras cannot accurately obtain visualized three-dimensional data information, but they can deduce the dynamic balance state of the impeller in three-dimensional space through two-dimensional data change characteristics.

[0070] Preferably, in one embodiment of the present invention, the impeller axial deflection index for the analysis period is obtained based on the position change of the centroid and the diameter change of the trajectory region formed by the trajectories of key points in the impeller edge images of consecutive frames during the analysis period. The specific method includes:

[0071] For any key point in the analysis period, the position of the key point in each frame of the impeller edge image is connected to the position in the adjacent frame of the impeller edge image to obtain the movement trajectory of the key point in the analysis period. The movement trajectory forms several closed regions, which are denoted as several trajectory regions. That is, due to the axial deflection of the key point during the rotation, multiple closed regions are formed, which are used as several trajectory regions. The centroid of each trajectory region and the length of the horizontal straight line passing through the centroid in the trajectory region are obtained as the horizontal diameter of each trajectory region.

[0072] Furthermore, for any key point and several trajectory regions during the analysis period, the distance between the centroids of any two trajectory regions is obtained. The horizontal diameters of all trajectory regions of the key point are arranged in ascending order to obtain the horizontal diameter sequence of the key point. The ratio of the latter horizontal diameter to the former horizontal diameter in the horizontal diameter sequence is obtained. The product of the mean of the ratios of all adjacent horizontal diameters in the horizontal diameter sequence and the standard deviation of the distance between the centroids of all pairwise trajectory regions of the key point is used as the axial deflection coefficient of the key point during the analysis period. The mean of the axial deflection coefficients of all key points during the analysis period is used as the impeller axial deflection index during the analysis period.

[0073] It should be noted that the larger the standard deviation of the distance between the centers of mass, the more likely the impeller is not centered during assembly, or there is an installation error between the shaft and the impeller. This can lead to axial back-and-forth "movement" during rotation. Large bearing clearance or loosening during operation can cause irregular axial movement of the impeller during rotation, resulting in an abnormally large drift. The larger the mean of the ratio of adjacent horizontal diameters, the more likely the transverse diameter of the movement trajectory is continuously increasing. This indicates that the radial instability of the impeller during rotation is intensifying, which is a typical manifestation of abnormal dynamic balance. In this case, the larger the impeller axial deflection index will be.

[0074] It should be further noted that if the error is caused by sampling error of the machine vision system, such as lens shake or image blur, it may also misjudge the lateral diameter of the impeller edge in the image as too large. In other words, it is necessary to eliminate the influence of machine vision sampling error.

[0075] Preferably, in one embodiment of the present invention, the impeller structure anomaly index for the analysis period is obtained by combining the grayscale changes of each blade region in adjacent frame impeller images, including the following specific method:

[0076] For any frame of impeller image in the analysis period, obtain the number of blade regions in the frame of impeller image, and the average of the average gray values ​​of pixels in each blade region. The product of the number of blade regions and the average of the average gray values ​​is used as the gray visual performance value of the frame of impeller image. Obtain the absolute value of the difference between the gray visual performance values ​​of any adjacent frame of impeller image in the analysis period.

[0077] Furthermore, based on the impeller axial deflection index during the analysis period and the mean of the absolute values ​​of the differences in grayscale visual performance values ​​of all adjacent frames of the impeller image during the analysis period, the impeller structure anomaly index during the analysis period is obtained. The impeller structure anomaly index is positively correlated with the impeller axial deflection index and negatively correlated with the mean of the absolute values ​​of the differences in grayscale visual performance values.

[0078] As an example, the ratio of the impeller axial deflection index during the analysis period to the mean of the absolute values ​​of the differences in grayscale visual representation values ​​of all adjacent frames of the impeller image during the analysis period is used as the impeller structure anomaly index for the analysis period.

[0079] It should be noted that the smaller the difference in grayscale visual performance values, the smaller the impact of machine vision error on the impeller axial deflection performance. The larger the impeller axial deflection index, the larger its structural anomaly index. By analyzing the impeller's operation process in space, the impeller structural anomaly index in the rotating state is obtained, which can more completely evaluate the dynamic balance state of the fan impeller.

[0080] It should be further explained that, based on the obtained impeller structure anomaly index, the non-ideal attitude deviation trend in the axial direction is identified, and combined with the impeller dynamic balance deviation characteristics, a multi-dimensional dynamic evaluation model is further established. This model not only considers the physical characteristics of the impeller mass distribution disturbance, but also integrates the visual change characteristics of the structural response under high-speed rotation, thereby making a more accurate and comprehensive judgment on the dynamic balance state and improving the credibility and operability of the evaluation results.

[0081] Preferably, in one embodiment of the present invention, the visual fusion dynamic balance index of the impeller during the analysis period is obtained based on the impeller dynamic balance deviation and impeller structural anomaly index during the analysis period. The specific method includes:

[0082] The sum of the inversely proportional normalized result of the impeller dynamic balance deviation during the analysis period and the inversely proportional normalized result of the impeller structural anomaly index during the analysis period is used as the visual fusion dynamic balance index of the impeller during the analysis period.

[0083] It should be noted that this embodiment adopts... The model is used to represent the inverse proportional relationship and for normalization processing. As input to the model, This represents an exponential function with the natural constant as the base. Implementers can set inverse proportional functions and normalization functions according to the actual situation.

[0084] It should be noted that the smaller the impeller dynamic balance deviation and impeller structural anomaly index, the larger the visual fusion dynamic balance index under the inverse proportion. The comprehensive output between the dynamic balance deviation dimension and the impeller's non-ideal attitude deviation trend in the axial direction can more comprehensively evaluate the dynamic balance state of the fan impeller, indicating that the dynamic balance state of the impeller is better.

[0085] Step S004: Adjust the initial dynamic balance threshold of the impeller based on the visual fusion dynamic balance index, obtain the adjusted dynamic balance threshold for the analysis period, and perform dynamic balance evaluation of the impeller.

[0086] Based on the equipment's design standards, manufacturer recommendations, and historical operating data, an initial dynamic balance threshold for the impeller is set. The product of this initial dynamic balance threshold and the visually fused dynamic balance index of the impeller during the analysis period is used as the adjusted dynamic balance threshold for that period. The dynamic balance of the axial flow fan impeller after the analysis period is then evaluated based on this adjusted threshold. The specific evaluation process follows existing technology in the field of impeller dynamic balance evaluation and will not be elaborated further in this embodiment. By dynamically adjusting the dynamic balance threshold, data-driven decision-making is emphasized, ensuring that the equipment's dynamic balance evaluation mechanism can adapt to different operating conditions, improving equipment reliability and safety. Simultaneously, equipment performance is optimized, maintenance costs are reduced, and overall operational efficiency is improved.

[0087] This concludes the embodiment.

[0088] It should be noted that, in this embodiment, in order to avoid the fraction being meaningless due to the denominator being 0, a hyperparameter is added to both the numerator and denominator for the ratio calculation. In this embodiment, the hyperparameter is described as 0.1, and will not be repeated in the above ratio calculation process.

[0089] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A machine vision-based method for evaluating the dynamic balance of axial flow fan impellers, characterized in that, The method includes the following steps: Acquire continuous frames of impeller images of a high-speed rotating axial flow fan, obtain impeller edge images for each frame corresponding to the analysis period and several moments within it, extract several key points from the impeller edge images, and record parameter data of several key operating conditions of the axial flow fan at each moment and the temperature at several locations in the impeller. Based on the trajectory smoothing of key points in the impeller edge images of consecutive frames during the analysis period, and the area distribution of the closed regions formed by the trajectories, the impeller load non-uniformity during the analysis period is obtained. Combining the changes in the area difference of different blades in the impeller edge images of consecutive frames, the dynamic instability index during the analysis period is obtained. Based on the changes in the dynamic instability index between adjacent time periods, and the differences in the temperature difference change trends at different positions of the impeller, combined with the changes in the parameter data fluctuations of each key operating condition, the impeller dynamic balance deviation during the analysis period is obtained. Based on the changes in the position of the centroid and the diameter of the trajectory region formed by the trajector edge images of key points in consecutive frames during the analysis period, the axial deflection index of the impeller during the analysis period is obtained; combined with the changes in the grayscale performance of each blade region in adjacent frames of impeller images, the impeller structural anomaly index of the analysis period is obtained; based on the impeller dynamic balance deviation and the impeller structural anomaly index during the analysis period, the visual fusion dynamic balance index of the impeller during the analysis period is obtained. The initial dynamic balance threshold of the impeller is adjusted based on the visual fusion dynamic balance index, the adjusted dynamic balance threshold for the analysis period is obtained, and the dynamic balance of the impeller is evaluated.

2. The machine vision-based dynamic balance evaluation method for axial flow fan impellers according to claim 1, characterized in that, The specific methods for obtaining the impeller load non-uniformity during the analysis period are as follows: The position of any key point in each frame of the impeller edge image during the analysis period is fitted to obtain the motion trajectory of the key point during the analysis period, and the closed area formed by the closed motion trajectory is obtained; the closed area is divided into a left area and a right area by a perpendicular line through the impeller center point, and the absolute value of the area difference between the left area and the right area is obtained as the closed area difference of the key point during the analysis period. The direction from the previous position to the next position of the key point in any adjacent frame of the impeller edge image during the analysis period is obtained as the motion direction of the key point in the previous frame of the adjacent frame during the analysis period. The motion trajectory of the key point in each frame during the analysis period is obtained. The standard deviation of the angle between the motion trajectories of the key point in all adjacent frames during the analysis period is multiplied by the difference in the closed area, which is used as the trajectory non-uniformity of the key point during the analysis period. The mean of the trajectory non-uniformity of all key points during the analysis period is taken as the impeller load non-uniformity during the analysis period.

3. The machine vision-based dynamic balance evaluation method for axial flow fan impellers according to claim 1, characterized in that, The specific methods for obtaining the dynamic instability index for the analysis period are as follows: For any frame of impeller edge image in the analysis period, obtain several blade regions in the frame of impeller edge image and obtain the area of ​​each blade region. Obtain the absolute value of the difference between the areas of any two blade regions. Use the mean of the absolute values ​​of the differences between the areas of all pairs of blade regions in the frame of impeller edge image as the blade area difference factor of the frame of impeller edge image. Obtain the absolute value of the difference in blade area difference factor between any two adjacent frames of impeller edge images during the analysis period. Multiply the mean of the absolute values ​​of the difference in blade area difference factor between all adjacent frames of impeller edge images during the analysis period by the impeller load non-uniformity during the analysis period, and use this product as the dynamic instability index for the analysis period.

4. The machine vision-based dynamic balance evaluation method for axial flow fan impellers according to claim 1, characterized in that, The specific methods for obtaining the impeller dynamic balance deviation during the analysis period are as follows: Based on the changes in the dynamic instability index between adjacent time periods and the differences in the temperature difference trends at different locations of the impeller, the impeller load trend index for the analysis period is obtained. Obtain the standard deviation of all parameter data for any key operating condition during the analysis period, obtain the standard deviation of all parameter data for the same key operating condition in the adjacent previous period of the analysis period, and use the ratio of the standard deviation of the key operating condition in the analysis period to the standard deviation of the same key operating condition in the adjacent previous period as the parameter fluctuation change factor for the key operating condition during the analysis period. Based on the impeller load trend index during the analysis period and the mean of the parameter fluctuation change factors for each key operating condition during the analysis period, the impeller dynamic balance deviation during the analysis period is obtained. The impeller dynamic balance deviation is positively correlated with the impeller load trend index and negatively correlated with the mean of the parameter fluctuation change factors for each key operating condition.

5. The machine vision-based dynamic balance evaluation method for axial flow fan impellers according to claim 4, characterized in that, The specific method for obtaining the impeller load trend index for the analysis period is as follows: For the temperature at several locations in the impeller at any given moment during the analysis period, obtain the absolute value of the temperature difference between any two locations, and take the average of the absolute values ​​of the temperature differences between all pairs of locations in the impeller at that moment as the average temperature difference at that moment. A coordinate system is constructed with time as the horizontal axis and temperature difference as the vertical axis. The average temperature difference at each moment in the analysis period is mapped to the coordinate system to obtain several data points. The data points are then fitted using the least squares method to obtain a fitted line and its slope. The slope is used as the temperature difference change factor for the analysis period. The ratio of the temperature difference change factor in the analysis period to the temperature difference change factor in the adjacent previous period is obtained, and the ratio of the dynamic instability index in the analysis period to the dynamic instability index in the adjacent previous period is obtained. The product of the two ratios is used as the impeller load trend index in the analysis period.

6. The machine vision-based dynamic balance evaluation method for axial flow fan impellers according to claim 2, characterized in that, The impeller axial deflection index for the analysis period is obtained using the following method: For any key point in the analysis period, the position of the impeller edge image in each frame is connected to the position in the impeller edge image of the adjacent frame to obtain the movement trajectory of the key point in the analysis period. The movement trajectory forms several closed regions, which are denoted as several trajectory regions. The centroid of each trajectory region and the length of the horizontal straight line passing through the centroid in the trajectory region are obtained as the horizontal diameter of each trajectory region. Based on the centroid distribution and horizontal diameter changes in the trajectory region of key points, the axial deflection coefficient of each key point during the analysis period is obtained. The average of the axial deflection coefficients of all key points during the analysis period is taken as the impeller axial deflection index for the analysis period.

7. The machine vision-based dynamic balance evaluation method for axial flow fan impellers according to claim 6, characterized in that, The axial deflection coefficients of each key point during the analysis period are obtained using the following method: For any key point in the analysis period, obtain the distance between the centroids of any two trajectory regions. Arrange the horizontal diameters of all trajectory regions of the key point in ascending order to obtain the horizontal diameter sequence of the key point. Obtain the ratio of the latter horizontal diameter to the former horizontal diameter in the horizontal diameter sequence. Multiply the mean of the ratios of all adjacent horizontal diameters in the horizontal diameter sequence with the standard deviation of the distance between the centroids of all pairwise trajectory regions of the key point as the axial deflection coefficient of the key point in the analysis period.

8. The machine vision-based dynamic balance evaluation method for axial flow fan impellers according to claim 3, characterized in that, The specific method for obtaining the impeller structure anomaly index for the analysis period is as follows: For any frame of impeller image in the analysis period, obtain the number of blade regions in the frame of impeller image, and the average of the average gray values ​​of pixels in each blade region. The product of the number of blade regions and the average of the average gray values ​​is used as the gray visual performance value of the frame of impeller image. Obtain the absolute value of the difference between the gray visual performance values ​​of any adjacent frame of impeller image in the analysis period. Based on the impeller axial deflection index during the analysis period and the mean of the absolute values ​​of the differences in grayscale visual representation values ​​of all adjacent frames of the impeller image during the analysis period, the impeller structure anomaly index for the analysis period is obtained. The impeller structure anomaly index is positively correlated with the impeller axial deflection index and negatively correlated with the mean of the absolute values ​​of the differences in grayscale visual representation values.

9. The machine vision-based dynamic balance evaluation method for axial flow fan impellers according to claim 1, characterized in that, The specific method for obtaining the visual fusion dynamic balance index of the impeller during the analysis period is as follows: The sum of the inversely proportional normalized result of the impeller dynamic balance deviation during the analysis period and the inversely proportional normalized result of the impeller structural anomaly index during the analysis period is used as the visual fusion dynamic balance index of the impeller during the analysis period.

10. The machine vision-based dynamic balance evaluation method for axial flow fan impellers according to claim 1, characterized in that, The specific method for obtaining the dynamic balance threshold for the analysis period is as follows: The product of the initial dynamic balance threshold of the impeller and the visual fusion dynamic balance index of the impeller during the analysis period is used as the adjustment dynamic balance threshold for the analysis period.

Images