CN116342898A - Method, device, equipment and storage medium for detecting wing-shaped icing area in icing wind tunnel - Google Patents
Method, device, equipment and storage medium for detecting wing-shaped icing area in icing wind tunnel Download PDFInfo
- Publication number
- CN116342898A CN116342898A CN202211227286.8A CN202211227286A CN116342898A CN 116342898 A CN116342898 A CN 116342898A CN 202211227286 A CN202211227286 A CN 202211227286A CN 116342898 A CN116342898 A CN 116342898A
- Authority
- CN
- China
- Prior art keywords
- icing
- wind tunnel
- region
- airfoil
- wing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000001514 detection method Methods 0.000 claims abstract description 48
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000012360 testing method Methods 0.000 claims abstract description 26
- 238000012545 processing Methods 0.000 claims abstract description 10
- 230000011218 segmentation Effects 0.000 claims abstract description 7
- 238000002474 experimental method Methods 0.000 claims description 19
- 239000011159 matrix material Substances 0.000 claims description 15
- 238000012937 correction Methods 0.000 claims description 13
- 238000004364 calculation method Methods 0.000 claims description 10
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000003384 imaging method Methods 0.000 claims description 7
- 238000013519 translation Methods 0.000 claims description 7
- 238000011176 pooling Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 3
- 238000000513 principal component analysis Methods 0.000 claims description 3
- 238000005096 rolling process Methods 0.000 claims description 3
- 230000000875 corresponding effect Effects 0.000 description 12
- 238000005286 illumination Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 5
- 238000002372 labelling Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000009434 installation Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 206010000369 Accident Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000013078 crystal Substances 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000036544 posture Effects 0.000 description 1
- 230000001681 protective effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4007—Scaling of whole images or parts thereof, e.g. expanding or contracting based on interpolation, e.g. bilinear interpolation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Aerodynamic Tests, Hydrodynamic Tests, Wind Tunnels, And Water Tanks (AREA)
Abstract
The invention relates to the technical field of icing detection, in particular to a method, a device, equipment and a storage medium for detecting wing-shaped icing areas in an icing wind tunnel. Firstly, shooting an experimental airfoil in an icing wind tunnel test section, acquiring an original image, and processing the original image to obtain training and testing image samples in an icing detection model training data set; training a semantic segmentation network for icing region detection by using the training data set; starting an icing wind tunnel, acquiring icing images of the wing profile in the test section in real time, inputting the icing images into a trained icing region detection model after processing, outputting position coordinates of a plurality of icing regions in a target image, finally judging the actual icing region to obtain icing regions on the wing profile, calculating to obtain the sum of the areas of all the icing regions on the wing profile, and combining a timer carried by a computer to calculate the icing speed of the wing profile under experimental conditions. The icing area of the wing profile is judged in real time through a machine vision means, and the icing area and the icing speed of the wing profile in the experimental process can be judged in real time with high accuracy.
Description
Technical Field
The present invention relates to the field of icing detection technologies, and in particular, to a method, an apparatus, a device, and a storage medium for detecting an airfoil icing region in an icing wind tunnel.
Background
From the early stage of aviation development, icing becomes one of the greatest threats for safe flight, and accounts for more than half of the world army and civil aircraft accidents. Icing on different parts of the aircraft can affect the safe flight of the aircraft. For example, icing of the wing, tail and control surfaces can dramatically degrade the aerodynamic performance and handling efficiency of the aircraft. In the design stage of the aircraft, the icing area and the icing type of each part of the aircraft can be obtained through an icing wind tunnel experiment, so that the anti-icing area of the aircraft is determined, and a basis is provided for the anti-icing design of the aircraft.
In the icing wind tunnel experiment, due to the severe experimental environment, scientific researchers usually only enter an icing wing section in an icing wind tunnel test section after the experiment is finished to analyze the icing wing section, record the icing type and icing quality, and lack corresponding calculation means for the icing area, icing speed and the like of the aircraft wing section when the experiment is carried out. Therefore, an image-based region detection algorithm can be introduced, the icing region of the surface of the airfoil is obtained in real time by using a deep learning method, and the icing speed of the airfoil in an icing wind tunnel experiment is calculated.
Because the experimental cost of the icing wind tunnel is too high, the obtained original images are fewer, and in the traditional deep learning model, the under fitting phenomenon possibly exists when the samples are fewer, namely the accuracy is lower when the model is inferred, and the ideal effect cannot be achieved. The UNet network proposed by Ciresan et al has the input training data of latches, so that the problem of fewer images can be effectively solved. Based on the problem that the UNet network training model is slow, the UNet training model is modified on the basis of UNet, the training speed is improved, meanwhile, the detection of an icing area can still keep high accuracy, and great help is provided for scientific researchers to the result analysis of the icing tunnel experiment.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention provides a method, a device, equipment and a storage medium for detecting wing-type icing areas in an icing wind tunnel.
In a first aspect, the present invention provides a method for detecting an airfoil icing region in an icing wind tunnel, where the technical scheme includes:
firstly, placing a calibration plate in an icing wind tunnel test section, ensuring that the imaging of the calibration plate in a camera is clear, recording a plurality of images of the calibration plate in different directions and angles by adjusting the position of the calibration plate, and obtaining an internal reference matrix of the camera, translation and rotation matrixes among the cameras and distortion coefficients of the cameras by calculation;
shooting an experimental airfoil by using an arranged camera to obtain an original image, carrying out distortion correction, defogging treatment and histogram equalization treatment on the image based on an image preprocessing algorithm, marking the outline of an icing region in an icing image on the surface of the airfoil by using a marking tool, and marking a corresponding label on the region to obtain training and testing image samples in an icing detection model training data set;
training a semantic segmentation network for detecting an icing region by using the training data set;
step four, starting an icing wind tunnel, acquiring an icing image of a wing type in a test section in real time, inputting the icing image into a trained icing region detection model after distortion correction, defogging and histogram equalization treatment, and outputting position coordinates (x i ,y i ) Wherein x is i 、y i Is the coordinates in the pixel coordinate system;
and fifthly, judging an actual icing region, converting the pixel coordinates of the output icing region into actual coordinates under a world coordinate system by using calibrated camera parameters, outputting the actual coordinates to obtain an icing region on the airfoil, calculating to obtain the sum of the areas of all the icing regions on the airfoil, and calculating to obtain the icing speed of the airfoil under the experimental condition by combining a timer of the computer.
Further, the implementation steps of the third step are as follows:
3.1, adopting a structure consisting of rolling and pooling and the like to finish feature extraction of the preprocessed airfoil surface image, repeating for a plurality of times to form a coding part of a network model, wherein the number of downsampling feature channels in each time is doubled;
step 3.2, in the decoding part, deconvoluting the obtained feature images, halving the number of feature channels, doubling the size of the feature images, splicing the deconvolution result and the result of the corresponding step of the encoding part, carrying out the deconvolution on the spliced feature images twice, repeating the deconvolution for a plurality of times, and finally converting the feature images into the result of the specific category number;
and 3.3, adopting a multi-supervision mode, inputting all the output of each part of the decoding stage into a loss function, carrying out back propagation calculation, and finally generating an icing region detection model, wherein the loss function is as follows:
where A is an a priori mask and B is a predictive mask.
Further, in the step 3.1, the primary component analysis is used to determine the initial value of the convolution kernel in each convolution layer, which specifically includes the steps of:
step 3.1.1, for a convolution kernel with the size of k multiplied by k, which is required by the coding part, extracting each input sample by using a patch with the size of k multiplied by k, and marking the extracted input samples into a row to form a data sample, and then, obtaining a covariance matrix of the data sample;
and 3.1.2, solving the first N eigenvectors of the covariance matrix for N convolution layers of each layer of the coding part, wherein the eigenvectors are 1 xkk, and respectively corresponding the values in the N eigenvectors to the values of M convolution kernels according to the row priority order.
Further, the fifth implementation step is as follows:
step 5.1, obtaining a proportionality coefficient k of a world coordinate system and a pixel coordinate system through a unit length on a pre-calibrated airfoil x 、k y ：
Wherein w is x 、w x The coordinate is the world coordinate system lower coordinate, x and y are the pixel coordinate system lower coordinate;
step 5.2, because the wind speed is higher in the icing wind tunnel, the newly-built ice in certain areas on the wing profile is easy to be blown away by wind, and the areas have no obvious influence on the wing profile, so that according to video display, a certain time is set as a threshold value according to a series of parameters of the icing wind tunnel experiment, and the ice still existing on the wing profile meeting the threshold value is considered to be actually frozen;
step 5.3, the icing zone coordinates (x i ,y i ) Converting into coordinates in a world coordinate system:
(w x ,w x )＝(k x x i ,k y y i )
step 5.4, finding the coordinate of the outermost periphery of the icing region through image binarization, and drawing the outline of the icing region by utilizing cubic spline interpolation;
step 5.5, calculating to obtain the sum S of the areas of all the icing areas of the airfoil:
wherein n is the number of airfoil icing areas, S i Icing area for individual zones;
step 5.6, because the invention judges the icing area based on the two-dimensional plane, the icing area increased in unit time is used as the icing speed v of the airfoil profile:
where Δs is the area of growth and t is time.
In a second aspect, the present invention provides a device for detecting an icing region of an airfoil in an icing wind tunnel, the device comprising:
the calibration module is used for calibrating the cameras installed in the test section to obtain an internal camera reference matrix, translation and rotation matrixes among the cameras and distortion coefficients of the cameras;
the acquisition module is used for acquiring an original image of wing type icing in the icing wind tunnel;
the preprocessing module is used for carrying out data processing on the original image, carrying out distortion correction on the original image by utilizing the distortion coefficient of the camera, and carrying out defogging and histogram equalization processing to obtain a data set after data preprocessing;
the training module is used for constructing an icing region detection model, constructing the icing region detection model by using the data set as input, optimizing model parameters and obtaining an inference model for icing region detection;
the detection module is used for acquiring an image in the wing section experiment in real time, inputting the image into the icing region detection model, and obtaining the icing region coordinates.
In a third aspect, the present invention provides an apparatus for detecting an icing region of an airfoil in an icing wind tunnel, the apparatus comprising:
one or more cameras and light sources for arranging in the icing wind tunnel test section and collecting airfoil experimental conditions;
a storage means for storing one or more programs and a data set required for the icing zone detection model;
the one or more processors are capable of implementing the wing type icing region detection method in an icing wind tunnel according to the first aspect of the present invention when the one or more programs are executed by the processor.
In a fourth aspect, the present invention provides a storage medium for detecting an airfoil icing region in an icing wind tunnel, where the storage medium is a computer readable medium, and one or more programs are stored, where the programs, when executed by a processor, can implement the method for detecting an airfoil icing region in an icing wind tunnel provided in the first aspect of the present invention.
Compared with the prior art, the invention has the beneficial effects that:
(1) The wing type icing region detection in the icing wind tunnel experiment is realized, and the icing area and the icing speed of the wing type in the experimental process can be judged in real time;
(2) The existing icing region detection model is improved, the training process of the model can be accelerated while the high accuracy rate of small sample training is maintained, and the training time is shortened;
(3) The wing icing region judgment method in the new icing wind tunnel experiment can judge the wing icing region in real time through a machine vision means.
Drawings
The invention is described in further detail below with reference to the attached drawings and detailed description:
FIG. 1 is a flow chart of the whole method of the invention;
FIG. 2 is a schematic flow chart of icing zone detection according to the present invention;
FIG. 3 is a graph showing the effect of judging icing areas in an icing wind tunnel.
Detailed Description
The invention is further described below with reference to the accompanying drawings:
in a first aspect, the present invention provides a method for detecting an airfoil icing region in an icing wind tunnel, where the technical scheme includes:
firstly, placing a calibration plate in an icing wind tunnel test section, ensuring that the imaging of the calibration plate in a camera is clear, recording a plurality of images of the calibration plate in different directions and angles by adjusting the position of the calibration plate, and obtaining an internal reference matrix of the camera, translation and rotation matrixes among the cameras and distortion coefficients of the cameras through calculation.
Recording a plurality of images of the calibration plate under different directions and angles, wherein the requirements are as follows: the number of the calibrated pictures at different positions, angles and postures shot by the camera is at least 3, preferably about 20.
The internal reference matrix of the camera and the translation and rotation matrix seen by each camera are obtained through calculation, and the requirements are as follows: converting each calibration picture into a gray level picture, extracting angular point information, optimizing sub-pixel angular points, and calculating by using an Opencv open source program to obtain internal and external parameters and distortion coefficients of the erected camera.
In a specific embodiment, 20 calibration plate photos with different poses are shot by using an erected camera, all the calibration photos are imported into an open source program, and internal and external parameters and distortion coefficients of the camera are calculated and derived.
Shooting an experimental airfoil by using an arranged camera to obtain an original image, carrying out distortion correction, defogging treatment and histogram equalization treatment on the image based on an image preprocessing algorithm, marking the outline of an icing region in an icing image on the surface of the airfoil by using a marking tool, and marking a corresponding label on the region to obtain training and testing image samples in an icing detection model training data set.
The method for shooting experimental airfoils by using arranged cameras acquires original images, and comprises the following steps: and determining the number and the placement positions of the cameras according to the size, the structure and the positions of the observation windows of the icing wind tunnel test section, determining the focal length of the cameras by measuring the distance between the cameras and the observation wing sections, and further selecting the proper cameras.
The camera should be a color camera, the resolution is at least 572×572×3, the lens should install the thermal insulation protective device, prevent the lens from fog and causing the imaging to be unclear in the experimental stage of the icing wind tunnel.
And shooting an experimental airfoil by using the installed camera, acquiring an original image, converting the camera imaging into a gray level image, determining the optimal installation positions and the number of the light sources in the experimental section by comparing the distribution of gray level values of the ice area and other areas of the airfoil, and selecting a proper light source installation mode in direct illumination, coaxial illumination, dark field illumination, diffuse illumination, backlight illumination and the like.
The selection of a suitable camera requires: the distortion correction is carried out on the obtained wing-shaped frozen image, and the requirements are as follows: the correction formulas of radial distortion and tangential distortion are:
x corrected_radial direction ＝x(1+k 1 r 2 +k 2 r 4 +k 3 r 6 )
y corrected_radial direction ＝y(1+k 1 r 2 +k 2 r 4 +k 3 r 6 )
x corrected_tangential direction ＝x+(2p 1 y+p 2 (r 2 +2x 2 )
y corrected_tangential direction ＝y+(2p 2 x+p 1 (r 2 +2y 2 )
Where x, y are the position coordinates of the distorted image and r is the distance of the point from the imaging center.
The histogram equalization processing is carried out on the image, and the requirements are as follows: the gray values with more pixel formats in the image (namely, the pixel values which play a main role in the picture) are widened, and the gray values with less pixels (namely, the gray values which do not play a main role in the picture) are merged, so that the contrast ratio is increased, the image is clear, and the purpose of enhancement is achieved.
Marking the outline of an icing region in an icing image on the surface of the airfoil by using a marking tool, and marking a corresponding label on the region, wherein the requirements are as follows: the semantic segmentation is essentially classification at the pixel level, which category the target pixel point belongs to is predicted, so that the segmentation labeling needs to label which pixels belong to which category, and an open-source labeling tool can be used for assisting in labeling the outline of the object.
In a specific embodiment, a full-color camera is selected for testing, the focal length of the camera is 10mm, the resolution is 1920 multiplied by 1080 multiplied by 3, the camera is placed outside an icing wind tunnel test section vertical to the side face of an experimental airfoil, shooting is carried out through glass, the lens is prevented from fogging due to low temperature, and clear imaging is ensured. A total of 16 LED lamps are selected as illumination light sources, and an installation mode of combining backlight and direct illumination is adopted, wherein each illumination mode has 8 LED lamps.
After the number and the positions of the cameras and the mounting positions of the light sources are determined, obtaining an image of the wing profile of the test section through the cameras, converting the image to obtain a gray level image, and finding out an icing area and a non-icing area on the surface of the wing profile, so that the follow-up steps can be carried out smoothly, taking pictures of wing profile icing conditions in about 1000 experiments by using the cameras as original images, and introducing a camera distortion coefficient to correct distortion.
And (3) for the original image subjected to distortion correction and defogging, introducing the original image into an Opencv open source program for histogram equalization, converting the histogram of the original image into uniform distribution, introducing the original image into LabelMe for pixel-level labeling of the outline of an icing region in an icing data set image, marking the outline label of the icing region, marking the icing label on the outline label, and marking Jing Biaoqian on the rest regions to obtain an icing region detection model training set and a verification set.
And thirdly, training a semantic segmentation network for detecting the icing region by using the training data set.
The semantic segmentation network for training icing region detection comprises:
3.1, adopting a structure consisting of rolling and pooling and the like to finish feature extraction of the preprocessed airfoil surface image, repeating for a plurality of times to form a coding part of a network model, wherein the number of downsampling feature channels in each time is doubled;
step 3.2, in the decoding part, deconvoluting the obtained feature images, halving the number of feature channels, doubling the size of the feature images, splicing the deconvolution result and the result of the corresponding step of the encoding part, carrying out the deconvolution on the spliced feature images twice, repeating the deconvolution for a plurality of times, and finally converting the feature images into the result of the specific category number;
and 3.3, adopting a multi-supervision mode, inputting all the output of each part of the decoding stage into a loss function, carrying out back propagation calculation, and finally generating an icing region detection model, wherein the loss function is as follows:
where A is an a priori mask and B is a predictive mask.
The step 3.1 of determining the initial value of the convolution kernel in each convolution layer by using principal component analysis comprises the following steps:
step 3.1.1, for a convolution kernel with the size of k multiplied by k, which is required by the coding part, extracting each input sample by using a patch with the size of k multiplied by k, and marking the extracted input samples into a row to form a data sample, and then, obtaining a covariance matrix of the data sample;
and 3.1.2, solving the first N eigenvectors of the covariance matrix for N convolution layers of each layer of the coding part, wherein the eigenvectors are 1 xkk, and respectively corresponding the values in the N eigenvectors to the values of M convolution kernels according to the row priority order.
The deconvolution result in the step 3.2 and the result of the corresponding step of the coding part are spliced together, and the requirements are as follows: if the feature map of the encoded portion is large in size, it is necessary to cut and then splice.
In a specific embodiment, the architecture of the coding part is composed of 4 repeated structures: 2 3 x 3 convolutional layers, a padding strategy of vaill, stride=1, a nonlinear ReLU layer and a 2 x 2 max pooling layer with stride of 2, the padding strategy being vaill; in the convolutional initial values of the convolutional layer extracted by principal component analysis, the selected patch size is 3 multiplied by 3, the first 2 eigenvectors of the mean square error matrix are taken, and the size of the eigenvectors is 1 multiplied by 9; similar to the coding layer, the deconvolution of the decoding part is also composed of four repeated structures, the deconvolution is firstly used before each repeated structure, the convolution kernel is 2×2 in size, after deconvolution, the characteristic diagram of the corresponding step of the coding part is cut and then spliced with the deconvolution result, the characteristic diagram after splicing is subjected to 3×3 convolutions for 2 times, the final layer is the convolution with the convolution kernel of 1×1, and the characteristic diagram of 64 channels is converted into an icing region and a background by using sigmoid; and carrying the output of each step of the decoding part into a loss function, carrying out back propagation calculation, and finally generating an icing region detection model.
And fourthly, starting an icing wind tunnel, acquiring an icing image of the wing type in the test section in real time, inputting the icing image into a trained icing region detection model after distortion correction, defogging and histogram equalization treatment, and outputting the position coordinates of the icing region in the target image.
The real-time acquisition of the icing image of the wing profile in the test section has the following requirements: and recording the video of the wing model icing in the experiment in the test section by using the erected camera, and processing the video frame by frame to obtain the wing model icing image.
In a specific embodiment, video of an airfoil experimental in an icing wind tunnel, which is shot by a camera, is subjected to distortion correction, defogging and histogram equalization, and is led into a UNet detection model which is trained in the fourth step, so that pixel coordinates of an icing region are obtained.
And fifthly, judging an actual icing region, converting the pixel coordinates of the output icing region into actual coordinates under a world coordinate system by using calibrated camera parameters, outputting the actual coordinates to obtain an icing region on the airfoil, calculating to obtain the sum of the areas of all the icing regions on the airfoil, and calculating to obtain the icing speed of the airfoil under the experimental condition by combining a timer of the computer.
The obtaining of the icing zone on the airfoil comprises the following steps:
step 5.1, obtaining a proportionality coefficient k of a world coordinate system and a pixel coordinate system through a unit length on a pre-calibrated airfoil x 、k y ：
Wherein w is x 、w x The coordinate is the world coordinate system lower coordinate, x and y are the pixel coordinate system lower coordinate;
step 5.2, because the wind speed is higher in the icing wind tunnel, the newly-built ice in certain areas on the wing profile is easy to be blown away by wind, and the areas have no obvious influence on the wing profile, so that according to video display, a certain time is set as a threshold value according to a series of parameters of the icing wind tunnel experiment, and the ice still existing on the wing profile meeting the threshold value is considered to be actually frozen;
step 5.3, the icing zone coordinates (x i ,y i ) Converting into coordinates in a world coordinate system:
(w x ,w x )＝(k x x i ,k y y i )
step 5.4, finding the coordinate of the outermost periphery of the icing region through image binarization, and drawing the outline of the icing region by utilizing cubic spline interpolation;
the sum of the areas of all icing areas on the airfoil is calculated, and the requirements are as follows:
wherein S is the sum of the areas of all icing areas on the airfoil, n is the number of the icing areas of the airfoil, S i An icing area for each zone.
The icing speed of the airfoil profile under the experimental condition is obtained through calculation, and the requirements are as follows: the icing area increased in unit time is used as the icing speed v of the airfoil, because the icing area is judged based on the two-dimensional plane:
where Δs is the area of growth and t is time.
In a specific embodiment, as shown in FIG. 3, the scaling factor k of the world coordinate system and the pixel coordinate system is obtained x 、k y ：
Due to thisIn the experiment, the wind speed is 50m/s, the MVD is 20um, the LWC is 1, and the experiment temperature is-15 ℃, so that the actual icing area of the airfoil surface is calculated to be 113.789mm after the airfoil surface is judged to be actually icing after 5s of ice crystals exist 2 Combining the icing area in the previous second video to obtain the real-time icing speed of 15.264mm on the surface of the airfoil 2 /s。
And (3) according to the camera parameters obtained in the step one, calculating to obtain the proportional relation between the world coordinate system and the pixel coordinate system, and carrying the pixel coordinate of the icing region in the step four to obtain the actual icing coordinate of the wing profile, namely obtaining the area of the wing profile icing region, the icing speed and the like.
In a second aspect, the present invention provides a device for detecting an icing region of an airfoil in an icing wind tunnel, the device comprising:
the calibration module is used for calibrating the cameras installed in the test section to obtain an internal camera reference matrix, translation and rotation matrixes among the cameras and distortion coefficients of the cameras;
the acquisition module is used for acquiring an original image of wing type icing in the icing wind tunnel;
the preprocessing module is used for carrying out data processing on the original image, carrying out distortion correction on the original image by utilizing the distortion coefficient of the camera, and carrying out defogging and histogram equalization processing to obtain a data set after data preprocessing;
the training module is used for constructing an icing region detection model, constructing the icing region detection model by using the data set as input, optimizing model parameters and obtaining an inference model for icing region detection;
the detection module is used for acquiring an image in the wing section experiment in real time, inputting the image into the icing region detection model, and obtaining the icing region coordinates.
In this embodiment, the airfoil icing region detection device in an icing wind tunnel can execute the airfoil icing region method in the icing wind tunnel provided by any embodiment of the invention, has all functional modules of the execution method, and can obtain corresponding effects.
In a third aspect, the present invention provides an apparatus for detecting an icing region of an airfoil in an icing wind tunnel, the apparatus comprising:
one or more cameras and light sources for arranging in the icing wind tunnel test section and collecting airfoil experimental conditions;
a storage means for storing one or more programs and a data set required for the icing zone detection model;
the one or more processors are capable of implementing the wing type icing region detection method in an icing wind tunnel according to the first aspect of the present invention when the one or more programs are executed by the processor.
In this example scenario, a computer device is provided, comprising: the device comprises a processor, a memory, an input device and an output device, wherein all the components are connected through a bus.
The memory is used for storing program software, a module corresponding to the icing region detection method in the example and a data set required by the model.
The processor needs to meet the memory and video memory requirements of the running program, and the icing region detection method is realized by executing a series of program instructions stored in the memory.
The input device mainly comprises a camera for experiments, a light source device for assisting in imaging, and other devices such as a mouse, a keyboard, a microphone and the like for receiving user instruction input.
The output device is used to effect presentation of the results of the iced areas in the examples described above, including but not limited to a display, a printer, and a microphone.
In a fourth aspect, the present invention provides a storage medium for detecting an airfoil icing region in an icing wind tunnel, where the storage medium is a computer readable medium, and one or more programs are stored, where the programs, when executed by a processor, can implement the method for detecting an airfoil icing region in an icing wind tunnel provided in the first aspect of the present invention.
The storage device, as a computer readable storage medium, includes electronic, magnetic, optical, infrared, etc., more specific examples include, but are not limited to, magnetic disks, hard disks, RAM, ROM, optical storage devices, magnetic storage devices, etc., or any combination thereof. In some examples, the memory may further include memory remotely located with respect to the processor, which may be connected to the processor device through a network connection including, but not limited to, the internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
In this embodiment, the method for detecting the icing area of the airfoil in the icing wind tunnel according to the first aspect of the present invention may be executed by a processor.
It should be noted that, for the apparatus, device and storage medium example schemes, since they are substantially similar to the method example schemes, the description is relatively simple, and the relevant points are referred to in the section of the method example schemes.
The above example schemes are merely exemplary example schemes employed to illustrate the principles of the present disclosure, however, the present disclosure is not limited thereto. Various modifications and improvements may be made by those skilled in the art without departing from the spirit and substance of the disclosure, and are also considered to be within the scope of the disclosure.
Claims (7)
1. The method for detecting the wing section icing region in the icing wind tunnel is characterized by comprising the following steps of:
firstly, placing a calibration plate in an icing wind tunnel test section, ensuring that the imaging of the calibration plate in a camera is clear, recording a plurality of images of the calibration plate in different directions and angles by adjusting the position of the calibration plate, and obtaining an internal reference matrix of the camera, translation and rotation matrixes among the cameras and distortion coefficients of the cameras by calculation;
shooting an experimental airfoil by using an arranged camera to obtain an original image, carrying out distortion correction, defogging treatment and histogram equalization treatment on the image based on an image preprocessing algorithm, marking the outline of an icing region in an icing image on the surface of the airfoil by using a marking tool, and marking a corresponding label on the region to obtain training and testing image samples in an icing detection model training data set;
training a semantic segmentation network for detecting an icing region by using the training data set;
step four, starting an icing wind tunnel, acquiring an icing image of a wing type in a test section in real time, inputting the icing image into a trained icing region detection model after distortion correction, defogging and histogram equalization treatment, and outputting position coordinates (x i ,y i ) Wherein x is i 、y i Is the coordinates in the pixel coordinate system;
and fifthly, judging an actual icing region, converting the pixel coordinates of the output icing region into actual coordinates under a world coordinate system by using calibrated camera parameters, outputting the actual coordinates to obtain an icing region on the airfoil, calculating to obtain the sum of the areas of all the icing regions on the airfoil, and calculating to obtain the icing speed of the airfoil under the experimental condition by combining a timer of the computer.
2. The method for detecting wing-type icing region in an icing wind tunnel according to claim 1, wherein the step three is specifically implemented as follows:
3.1, adopting a structure consisting of rolling and pooling and the like to finish feature extraction of the preprocessed airfoil surface image, repeating for a plurality of times to form a coding part of a network model, wherein the number of downsampling feature channels in each time is doubled;
step 3.2, in the decoding part, deconvoluting the obtained feature images, halving the number of feature channels, doubling the size of the feature images, splicing the deconvolution result and the result of the corresponding step of the encoding part, carrying out the deconvolution on the spliced feature images twice, repeating the deconvolution for a plurality of times, and finally converting the feature images into the result of the specific category number;
and 3.3, adopting a multi-supervision mode, inputting all the output of each part of the decoding stage into a loss function, carrying out back propagation calculation, and finally generating an icing region detection model, wherein the loss function is as follows:
where A is an a priori mask and B is a predictive mask.
3. The method for detecting wing-type icing zone in an icing wind tunnel according to claim 1, wherein the step 3.1 of determining the initial value of the convolution kernel in each convolution layer by using principal component analysis comprises the following specific steps:
step 3.1.1, for a convolution kernel with the size of k multiplied by k, which is required by the coding part, extracting each input sample by using a patch with the size of k multiplied by k, and marking the extracted input samples into a row to form a data sample, and then, obtaining a covariance matrix of the data sample;
and 3.1.2, solving the first N eigenvectors of the covariance matrix for N convolution layers of each layer of the coding part, wherein the eigenvectors are 1 xkk, and respectively corresponding the values in the N eigenvectors to the values of M convolution kernels according to the row priority order.
4. The method for detecting wing-type icing region in an icing wind tunnel according to claim 1, wherein the fifth implementation step comprises the following steps:
step 5.1, obtaining a proportionality coefficient k of a world coordinate system and a pixel coordinate system through a unit length on a pre-calibrated airfoil x 、k y ：
Wherein w is x 、w x The coordinate is the world coordinate system lower coordinate, x and y are the pixel coordinate system lower coordinate;
step 5.2, because the wind speed is higher in the icing wind tunnel, the newly-built ice in certain areas on the wing profile is easy to be blown away by wind, and the areas have no obvious influence on the wing profile, so that according to video display, a certain time is set as a threshold value according to a series of parameters of the icing wind tunnel experiment, and the ice still existing on the wing profile meeting the threshold value is considered to be actually frozen;
step 5.3, the icing zone coordinates (x i ,y i ) Converting into coordinates in a world coordinate system:
(w x ,w x )＝(k x x i ,k y y i )
step 5.4, finding the coordinate of the outermost periphery of the icing region through image binarization, and drawing the outline of the icing region by utilizing cubic spline interpolation;
step 5.5, calculating to obtain the sum S of the areas of all the icing areas of the airfoil:
wherein n is the number of airfoil icing areas, S i Icing area for individual zones;
step 5.6, because the invention judges the icing area based on the two-dimensional plane, the icing area increased in unit time is used as the icing speed v of the airfoil profile:
where Δs is the area of growth and t is time.
5. An icing zone detection device for a wing type in an icing wind tunnel, which is characterized by comprising:
the calibration module is used for calibrating the cameras installed in the test section to obtain an internal camera reference matrix, translation and rotation matrixes among the cameras and distortion coefficients of the cameras;
the acquisition module is used for acquiring an original image of wing type icing in the icing wind tunnel;
the preprocessing module is used for carrying out data processing on the original image, carrying out distortion correction on the original image by utilizing the distortion coefficient of the camera, and carrying out defogging and histogram equalization processing to obtain a data set after data preprocessing;
the training module is used for constructing an icing region detection model, constructing the icing region detection model by using the data set as input, optimizing model parameters and obtaining an inference model for icing region detection;
the detection module is used for acquiring an image in the wing section experiment in real time, inputting the image into the icing region detection model, and obtaining the icing region coordinates.
6. An airfoil icing region check out test set in an icing wind tunnel, characterized in that: comprising the following steps:
one or more cameras and light sources for arranging in the icing wind tunnel test section and collecting airfoil experimental conditions;
a storage means for storing one or more programs and a data set required for the icing zone detection model;
the one or more processors are capable of implementing the wing type icing region detection method in an icing wind tunnel according to the first aspect of the present invention when the one or more programs are executed by the processor.
7. The utility model provides a wing type icing area detects storage medium in icing wind-tunnel which characterized in that: the storage medium is a computer readable medium, and stores one or more programs, which when executed by a processor, can implement the method for detecting wing type icing area in an icing wind tunnel according to the first aspect of the present invention.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211227286.8A CN116342898A (en) | 2022-10-09 | 2022-10-09 | Method, device, equipment and storage medium for detecting wing-shaped icing area in icing wind tunnel |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211227286.8A CN116342898A (en) | 2022-10-09 | 2022-10-09 | Method, device, equipment and storage medium for detecting wing-shaped icing area in icing wind tunnel |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116342898A true CN116342898A (en) | 2023-06-27 |
Family
ID=86886346
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211227286.8A Pending CN116342898A (en) | 2022-10-09 | 2022-10-09 | Method, device, equipment and storage medium for detecting wing-shaped icing area in icing wind tunnel |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116342898A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117061859A (en) * | 2023-10-12 | 2023-11-14 | 中国空气动力研究与发展中心低速空气动力研究所 | Icing wind tunnel test camera monitoring system and method |
-
2022
- 2022-10-09 CN CN202211227286.8A patent/CN116342898A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117061859A (en) * | 2023-10-12 | 2023-11-14 | 中国空气动力研究与发展中心低速空气动力研究所 | Icing wind tunnel test camera monitoring system and method |
| CN117061859B (en) * | 2023-10-12 | 2023-12-22 | 中国空气动力研究与发展中心低速空气动力研究所 | Icing wind tunnel test camera monitoring system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107945204B (en) | Pixel-level image matting method based on generation countermeasure network | |
| CN104217404A (en) | Video image sharpness processing method in fog and haze day and device thereof | |
| CN112580407B (en) | Space target part identification method based on lightweight neural network model | |
| CN113506214B (en) | Multi-path video image stitching method | |
| CN111680705B (en) | MB-SSD method and MB-SSD feature extraction network suitable for target detection | |
| CN111489330B (en) | Weak and small target detection method based on multi-source information fusion | |
| CN112258490A (en) | Low-emissivity coating intelligent damage detection method based on optical and infrared image fusion | |
| CN113495575B (en) | Unmanned aerial vehicle autonomous landing visual guidance method based on attention mechanism | |
| CN115240089A (en) | Vehicle detection method of aerial remote sensing image | |
| CN116342898A (en) | Method, device, equipment and storage medium for detecting wing-shaped icing area in icing wind tunnel | |
| CN111461006B (en) | Optical remote sensing image tower position detection method based on deep migration learning | |
| CN114881931A (en) | Photovoltaic panel defect detection method, system and equipment based on double-light fusion and computer readable storage medium | |
| CN112508814B (en) | Image tone restoration type defogging enhancement method based on unmanned aerial vehicle at low altitude visual angle | |
| CN114627044A (en) | Solar photovoltaic module hot spot detection method based on deep learning | |
| CN113378672A (en) | Multi-target detection method for defects of power transmission line based on improved YOLOv3 | |
| CN117372764A (en) | Non-cooperative target detection method in low-light environment | |
| CN116993975A (en) | Panoramic camera semantic segmentation method based on deep learning unsupervised field adaptation | |
| CN116740572A (en) | Marine vessel target detection method and system based on improved YOLOX | |
| CN116363610A (en) | Improved YOLOv 5-based aerial vehicle rotating target detection method | |
| CN113763261B (en) | Real-time detection method for far small target under sea fog weather condition | |
| CN115100680A (en) | Pedestrian detection method based on multi-source image fusion | |
| CN114998801A (en) | Forest fire smoke video detection method based on contrast self-supervision learning network | |
| Chen et al. | A novel monocular calibration method for underwater vision measurement | |
| CN113506230A (en) | Photovoltaic power station aerial image dodging processing method based on machine vision | |
| CN112434606A (en) | Double-light-integration-based solar cell panel RGB image recognition method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |