CN113884538A - Infrared thermal image detection method for micro defects in large wind turbine blade - Google Patents

Abstract

The invention belongs to the technical field of infrared detection, and particularly relates to an infrared thermography detection method for micro defects in a large wind turbine blade. The invention is suitable for large-area detection, and the defect information can be automatically identified in the blade heating process. The method comprises the following steps: s1, acquiring the linear distance between the thermal imager and the surface of the sample, measuring the corresponding length and width of the shooting area of the thermal imager at different distances, and respectively establishing a relation function; s2, continuously irradiating the front surface of the wind turbine blade by using a heat source; s3, continuously collecting infrared thermographs; s4, representing each frame of thermal image as a matrix; S5-S12, extracting dynamic defect features through space dimension reduction to enable the defect features to be secondarily imaged and enhanced; s13, obtaining the actual distance between the thermal imager and the blade, and substituting the actual distance into the calculation formula in S1 to obtain the length and the width; s14, calculating the average value of the element values outside the region, selecting the element with the fastest temperature rise in the region, and extracting the temperature sequence of the element; and S15, obtaining the detection result of the defect depth.

Description

Infrared thermal image detection method for micro defects in large wind turbine blade

Technical Field

The invention belongs to the technical field of infrared detection, and particularly relates to an infrared thermography detection method for micro defects in a large wind turbine blade.

Background

Wind power generation is an important component of new energy in China, and as a core device of wind power generation, long-term and stable operation of a wind turbine is important to energy safety production. The blades are the main load-bearing equipment of the wind turbine, which is expensive to purchase and maintain, and once damaged, it will cause huge shutdown losses and serious safety accidents. With the upsizing of wind turbines, the problems of high operation and maintenance costs and safety risks caused by the problem of blade mass become increasingly serious. Therefore, nondestructive testing for wind turbine blades has become the key content of wind turbine operation and maintenance at present.

The blade material inevitably has process defects of layering, pores, folds and the like in the production and manufacturing process. These defects can induce damage initiation and propagation under the long-term effects of complex loads, ultimately leading to blade failure and even early failure. Generally, the smaller the defect size, the more difficult the detection, especially under strong interference conditions, the easier the detection omission and false detection. The method for quantitatively identifying the tiny defects of the blade by utilizing the nondestructive testing method is important for evaluating the state of the blade, realizing early maintenance and predicting the service life of the blade. The infrared thermal imaging technology is widely applied to blade defect detection due to the advantages of non-contact, safety, no damage, easy operation, imaging, full-field detection and the like. However, the existing thermal infrared imaging technology still has certain limitations when detecting the small defects in the blades, and is specifically embodied as follows:

firstly, the interference resistance is poor: in the field detection process, due to the action of thermal interference such as non-ideal heat sources, anisotropy of blade materials, complex shapes, environmental thermal noise and the like and the heat flow around the defects, the micro defects in the blades are easily subjected to false detection and missed detection. Particularly, when large-area detection is carried out, the defective pixel ratio is small, and the identification difficulty is higher. The existing methods generally improve the quality of infrared heat maps frame by frame to improve the definition of defects, but the static processing methods have poor application effect in the face of strong interference with obvious change rate and can seriously lose effective information of the defects. Therefore, there is a need to develop an algorithm that can effectively identify minute defects inside the blade under strong interference conditions.

Secondly, the precision is not ideal: the existing detection method is mostly established based on a one-dimensional heat conduction model, and the anisotropy of a blade material and the three-dimensional effect of heat flow are not considered, so that the large-size defect can only be accurately detected. The smaller the defect size, the more significant the detection bias due to model limitations. Therefore, it is necessary to accurately detect the minute defects based on the three-dimensional thermal conduction model of the anisotropic material.

Thirdly, the efficiency is low: to ensure that all internal blade defects are covered by heat, existing inspection methods typically require long periods of heating or cooling of the blade, with fixed heating times. Furthermore, the data analysis process of the existing inspection method needs to be started after the heating or cooling of the blade is finished. These methods are time consuming, and especially when large area inspection is performed, the problem is more prominent because of the small proportion of the micro-defect pixels. Therefore, it is necessary to optimize the detection process and improve the detection efficiency.

In addition, the existing detection methods also include:

1. the infrared thermograph, particularly a defect area has a serious ghost under the combined action of factors such as non-ideal heat sources, anisotropy of blade materials, environmental thermal interference, three-dimensional heat flow around defects and the like, so that the identifiability of tiny defects is poor. There is a need to develop a heat map processing algorithm, which can effectively extract and clearly display the defect features under the condition of strong interference, so that the identifiability of the tiny defects and the detection precision of the geometric information are improved.

2. Most of the existing thermal infrared imaging technologies are based on one-dimensional heat conduction models, and due to the fact that anisotropy of materials and three-dimensional effects of heat flow cannot be considered, accurate assessment on depth of micro defects in the blades is difficult to achieve. There is a need to develop a defect depth evaluation algorithm based on a three-dimensional heat conduction model to realize accurate detection of the depth of a micro defect.

3. Most of the existing infrared thermal imaging methods adopt a detection mode of heating before analysis. This method requires long heating times to achieve sufficient deep defect detection capability, is time consuming, inefficient to detect, and relies on manual identification of defect contours. The smaller the defect size, the longer the inspection time, and the lower the inspection efficiency. There is a need to design a new detection strategy to realize detection while heating and to accomplish quantitative and automatic defect identification in the shortest time.

Disclosure of Invention

Aiming at the defects in the prior art, the invention provides an infrared thermal image detection method for micro defects in the large wind turbine blade. The method completes the feature extraction and contrast enhancement of the defects by using a principal component analysis method containing a variable-pitch contrast enhancement strategy, solves the problem of low micro-defect identifiability caused by strong thermal interference, small pixel proportion, thermal image ghost and the like, and improves the identification capability of the micro-defects of the leaves. The heat map processing algorithm is used for inhibiting thermal interference and virtual images around the defects, the problem that the depth of the micro defects is difficult to detect is solved by using the anisotropic material three-dimensional heat conduction model, and accurate quantitative detection of main information such as the position, the shape, the size and the depth of the micro defects is realized. The method is suitable for large-area detection, unnecessary heating and cooling time can be effectively avoided by a detection mode of heating and analyzing simultaneously, and the defect information can be automatically identified in the blade heating process.

In order to achieve the purpose, the invention adopts the following technical scheme that the method comprises the following steps:

s1, acquiring the linear distance d of the thermal imager from the surface of the sample piece (under laboratory conditions)_sMeasuring the length l and the width w corresponding to the shooting area of the thermal imager at different distances, and respectively establishing d by using a least square method_sAnd l, d_sAnd W is expressed as (H, W) ═ f_a(d_s)；

S2, continuously irradiating the front surface of the wind turbine blade by using a heat source;

s3, continuously collecting infrared thermal images (with infrared thermal imager at 1Hz sampling frequency), wherein the total sampling frame number is F_r；

S4, representing each frame as a matrix

N represents the total number of columns, M represents the total number of rows, a_μ,νRepresenting the actual temperature of each shooting pixel point, wherein mu represents the row number of a, and ν represents the column number of a respectively, and μ is 1,2,. M, ν is 1,2,. N;

S5-S12, extracting dynamic defect features through space dimensionality reduction, and secondarily imaging and enhancing the defect features while keeping information integrity by using a variable-spacing contrast enhancement algorithm;

s13, obtaining the actual distance d between the thermal imager and the blade (by using a laser range finder)_sAnd substituted into the calculation formula in S1 to obtain the length L and the width w, and then the length L is L × N according to the formula L_L/320 obtaining the defect length L and W × N according to the formula_L240 obtaining the defect width W;

s14, calculation A_iMean value T of the values of the elements outside the region E_L,iThen, the element with the fastest temperature rise in the region E is selected, and the temperature sequence T of the element is extracted_B,iCalculating a difference T between the two_e,i＝T_B,i-T_L,iWhen T is recorded_e,i/T_L,iHeating time t corresponding to the number of frames i when the time is 0.05_c；

And S15, obtaining the detection result of the defect depth.

Further, in S5: defining a variable pitch contrast enhancement matrix I, and setting I (mu, v) to k_a(μ)×a_μ,ν ²+k_b(μ)×a_μ,ν+k_c(μ)，k_a、k_b、k_cIs a coefficient related to the number of rows μ;

in S6: a. the_iOptional three points in each row are used as iteration points and substituted into I (mu, v) to obtain a fitting value I_t(ii) a Let λ remaining elements in the row except the selected three points be greater than I_tBeta remaining elements are less than the value, and the polynomial lambda^3/2-1.5λβ^1/2+0.75λ^1/2β-0.125β³Is smaller than that when other elements are selected as iteration points, k corresponding to the iteration point is determined_a,k_b,k_cThe fitting coefficient of the line is calculated line by line to obtain the fitting coefficient of each linek_a(μ),k_b(μ),k_c(μ)；

In S7: a is to be_iAll the elements are substituted into I in sequence_i(mu, v) and subtracted to obtain a new matrix B_iI.e. B_i＝A_i-I_i；

In S8: b is to be_iRewriting to a single column vector T_i＝(b₁₁…b_M1…b_1N…b_MN)^TThen will be F together_rT of Frames_iAre combined into a matrix T ═ (T)₁,T₂,…,T_Fr)；

In S9: covariance matrix T of calculation matrix T_xIs a

In S10: let matrix T ═ T_x×T_x ^TCalculating the eigenvector V of T' when the singular value is maximum, and rearranging the elements in the vector V according to M rows and N columns to obtain a matrix C, namely

In S11: extracting the maximum value C of the matrix C_maxAnd minimum value C_minDefining a new matrix D and letting D_μ,ν＝(C_x,y-C_min)×255/(C_max-C_min)；

In S12: defining a closed area E in the matrix D, enabling elements in the area E to meet the sum of the maximum value and the sum of the maximum values of 32, recording coordinates of all elements in the area E, calculating the maximum row number difference and the maximum column number difference between any elements in the area E, and respectively recording the maximum row number difference and the maximum column number difference as N_LAnd N_W。

Further, in S15, let ρ be the blade material density, c be the heat capacity, and K be_pIs the x-y in-plane thermal conductivity, K, of the blade material_zFor the thermal conductivity of the blade material in the z direction (thickness direction), let

H＝h_r/K_z，

α＝K/(ρ·c)，α_z＝K_z/(ρ·c)，

The length of the blade is h₁Width of h₂Coefficient of K_X1＝2h₁l₁-X-ll₁-2nh₁l₁+2nll₁，K_X2＝X-ll₁+2nh₁l₁+2mll₁，K_Y1＝2h₂l₁-Y-wl₁-2mh₂l₁+2mwl₁，K_Y2＝Y-bl₁+2mh₂l₁+2mwl₁，K_Z1＝Z+dl₂+2pdl₂，K_Z2＝-Z+dl₂+2pdl₂；

Defining x-direction contrast

And contrast in the y-direction

And also the contrast in the thickness direction

In the formula, m, n, r and p are summation parameters, and the polynomial C is substituted by taking d as 0.0,0.1,0.2_x×C_y×C_zThe value of d, where the polynomial value is closest to 0.05, is the depth of defect value.

Specifically, in S1: adjusting the linear distances between the thermal imager and the surface of the sample to be d_sRespectively detecting the corresponding length l and width w of the shooting area of the thermal imager under different distances by using a graduated scale and respectively establishing d by using a least square method, wherein the length l and the width w are 0.1m,0.2m, … m and 2m_sAnd l, d_sAnd w, the relation function of the invention is specifically that l is 0.46 × d_sAnd w is 0.24 × d_s。

Compared with the prior art, the invention has the beneficial effects.

The invention has the following anti-interference functions: the defect dynamic characteristics are extracted through space dimensionality reduction by utilizing the step difference between a thermal interference source and a defect temperature field, and the defect characteristics are secondarily imaged while information integrity is kept by utilizing a variable pitch contrast enhancement algorithm, so that defect identifiability is improved, and effective identification of millimeter-scale micro defects in the large wind turbine blade under the condition of strong thermal interference is finally realized.

Drawings

The invention is further described with reference to the following figures and detailed description. The scope of the invention is not limited to the following expressions.

FIG. 1 is a flow chart of blade micro-defect detection.

FIG. 2 is a schematic view of a sample blade containing a defect.

FIG. 3 is an infrared thermography image under strong interference conditions.

FIG. 4 is an infrared thermographic image processed by a thermographic processing algorithm.

FIG. 5 is a schematic diagram of a defect and sound area sampling setup.

Figure 6 shows (a) a x b is 10.0mm x 10.0mm,

curve line.

Figure 7 shows (b) a x b is 20.0mm x 20.0mm,

curve line.

Fig. 8 shows (c) a × b is 30.0mm × 30.0mm,

curve line.

Detailed Description

The whole detection process is shown in the attached figure 1.

Step S1: under the condition of a laboratory, acquiring the linear distance d between the thermal imager and the surface of the sample_sFunction f relating to length H and width W of the region to be imaged_aTo obtain H ═ 0.46d_s,W＝0.32d_s；

Manufacturing a blade composite material sample, and processing 12 flat-bottom square hole defects with different sizes and depths on the back of the sample, wherein the serial numbers are d₁,d₂,...,d₁₂The appearance of the sample is shown in figure 2. The defect conditions are shown in table 1:

TABLE 1 Defect information Table

Steps S2-S4: 2 halogen lamps which are symmetrically arranged and have the power of 1kW are used as excitation sources to heat the front surface of the sample piece with the defects, the ambient temperature is 25.0 ℃, the heating distance is 0.5m, and the heating time is uniformly set to 600 s. And after the single thermal loading is finished, cooling the sample piece to the ambient temperature and performing repeated tests after ensuring that the initial temperature field on the surface of the sample piece has no singularity, so as to obtain 30 groups of sample data. The temperature change of the surface of the sample piece is captured by an NEC R300 thermal imager with the resolution of 320 multiplied by 240 and the thermal resolution of 30mK, and the measuring range is fixed to be 22.0-60.0 ℃. The heat map data is transmitted to a computer and the heat map data processing software completes the work of temperature conversion, temperature rise curve analysis and the like. The sampling time interval is 1s and the number of sample frames is 1000.

Steps S5-S12: and extracting dynamic characteristics of the defect through space dimension reduction, and secondarily imaging and enhancing the characteristics of the defect by using a variable-pitch contrast enhancement algorithm while maintaining the integrity of information. The effect after treatment is shown in figure 4.

Step S13, detecting distance d_s0.8m, the lengths L and widths W of the respective defects were obtained as shown in Table 2：

TABLE 2 statistical table of defect size measurement results

Therefore, the false image before the heat map processing covers the defect characteristics, so that the defect cannot be accurately identified. After the image processing algorithm is used for processing, the defect identifiability is remarkably improved, the identification precision of the defect size is not more than 12%, and the precise identification of the tiny defect with the size within 15mm multiplied by 15mm is realized.

Table 3 lists the automatic identification times for each defect. It can be seen that each deep defect with a depth of 10mm can be effectively identified within 160s, and the identification content includes an assessment of the depth of the defect.

TABLE 3 automatic identification of defects time statistics table

Step S14: as shown in figure 5, the pixel point with the fastest temperature rise in the defect area is taken as a defect sampling point, the temperature of each defect is obtained and is subtracted from the ambient temperature measured by the thermal imager, and the excess temperature T of each defect is obtained₁,T₂,…,T₁₂. Taking a sample area (excluding the clamp area) outside the solid frame as a sound area to obtain the excess temperature T of the sound area_s(t)。

For example, defect d corresponding to the 100 th s heating₁The temperature is 30.0 ℃, the average temperature of the sound area is 28.0 ℃, and the defect excess temperature is T_e│_t＝100s30.0-28.0 deg.C-2.0 deg.C. Using formula C_r＝T_e/T_sCalculating relative temperature rise C of defect_r│_t＝100s＝T_e/T_s0.07. FIGS. 6-8 show C for all defects_r(t) curve.

Step S15: the size of each defect and the separation time t_cSubstituting the evaluation formula of the step S15 to obtain the defect depthDegree of detection result d_{p_3D}Let the true depth of the defect be d_rAnd obtaining the depth detection deviation eta \u_3D＝100×│(d_{p_3D}-d_r)│/d_r% of the total weight of the composition. The test results are collated in Table 4.

TABLE 4 Defect depth detection results

As shown in Table 4, the detection deviation of the defect depth was 6.0% or more but not more than η_{1_3D}Less than or equal to 10.0 percent and average detection deviation

The depth of the micro defects in the blade can be accurately detected.

The invention also has the following advantages:

and (3) accurate detection: the method has the advantages that the secondary imaging defect dynamic characteristics are adopted, the variable-interval contrast enhancement is carried out, the virtual image around the micro defect is effectively inhibited, the detection precision of the geometric information of the micro defect is improved, the defect depth is quantitatively evaluated based on the three-dimensional anisotropic material heat conduction model, and the precise detection of the micro defect depth is realized;

and (3) fast identification: the defect thermal contrast at the early stage of heating is used as an index to realize the automatic adjustment of the heating time, and the data analysis and the heating process are synchronously carried out, so that the detection time is effectively saved, the method is also suitable for automatic and large-area detection, and the heat map processing algorithm can further shorten the micro defect identification time.

The application range is wide: the algorithm is not limited by blade material, size and defect conditions, and provides sufficient heat to cover any depth, any size of defect by continuous heating.

It should be understood that the detailed description of the present invention is only for illustrating the present invention and is not limited by the technical solutions described in the embodiments of the present invention, and those skilled in the art should understand that the present invention can be modified or substituted equally to achieve the same technical effects; as long as the use requirements are met, the method is within the protection scope of the invention.

Claims (3)

1. The infrared thermal image detection method for the tiny defects in the large wind turbine blade is characterized by comprising the following steps: the method comprises the following steps:

s1, acquiring the linear distance d between the thermal imager and the surface of the sample piece_sMeasuring the length l and the width w corresponding to the shooting area of the thermal imager at different distances, and respectively establishing d by using a least square method_sAnd l, d_sAnd W is expressed as (H, W) ═ f_a(d_s)；

S2, continuously irradiating the front surface of the wind turbine blade by using a heat source;

s3, continuously collecting infrared thermography, wherein the total sampling frame number is F_r；

S4, representing each frame as a matrix

N represents the total number of columns, M represents the total number of rows, a_μ,νRepresenting the actual temperature of each shooting pixel point, wherein mu represents the row number of a, and ν represents the column number of a respectively, and μ is 1,2,. M, ν is 1,2,. N;

S5-S12, extracting dynamic defect features through space dimensionality reduction, and secondarily imaging and enhancing the defect features while keeping information integrity by using a variable-spacing contrast enhancement algorithm;

s13, obtaining the actual distance d between the thermal imager and the blade_sAnd substituted into the calculation formula in S1 to obtain the length L and the width w, and then the length L is L × N according to the formula L_L/320 obtaining the defect length L and W × N according to the formula_L240 obtaining the defect width W;

s14, calculation A_iMean value T of the values of the elements outside the region E_L,iThen, the element with the fastest temperature rise in the region E is selected, and the temperature sequence T of the element is extracted_B,iCalculating a difference T between the two_e,i＝T_B,i-T_L,iWhen T is recorded_e,i/T_L,iHeating time t corresponding to the number of frames i when the time is 0.05_c；

And S15, obtaining the detection result of the defect depth.

2. The infrared thermography detection method for the micro defects in the blades of the large wind turbine according to claim 1, wherein the infrared thermography detection method comprises the following steps: in S5: defining a variable pitch contrast enhancement matrix I, and setting I (mu, v) to k_a(μ)×a_μ,ν ²+k_b(μ)×a_μ,ν+k_c(μ)，k_a、k_b、k_cIs a coefficient related to the number of rows μ;

in S6: a. the_iOptional three points in each row are used as iteration points and substituted into I (mu, v) to obtain a fitting value I_t(ii) a Let λ remaining elements in the row except the selected three points be greater than I_tBeta remaining elements are less than the value, and the polynomial lambda^3/2-1.5λβ^1/2+0.75λ^1/2β-0.125β³Is smaller than that when other elements are selected as iteration points, k corresponding to the iteration point is determined_a,k_b,k_cIs the fitting coefficient of the line, and then calculates line by line and obtains the fitting coefficient k of each line_a(μ),k_b(μ),k_c(μ)；

In S7: a is to be_iAll the elements are substituted into I in sequence_i(mu, v) and subtracted to obtain a new matrix B_iI.e. B_i＝A_i-I_i；

In S8: b is to be_iRewriting to a single column vector T_i＝(b₁₁…b_M1…b_1N…b_MN)^TThen, the T of Fr frame_iAre combined into a matrix T ═ (T)₁,T₂,…,T_Fr)；

In S9: covariance matrix T of calculation matrix T_xIs a

In S10: let matrix T ═ T_x×T_x ^TCalculating the eigenvector V of T' when the singular value is maximum, and rearranging the elements in the vector V according to M rows and N columns to obtain a matrix C, namely

In S11: extracting the maximum value C of the matrix C_maxAnd minimum value C_minDefining a new matrix D and letting D_μ,ν＝(C_x,y-C_min)×255/(C_max-C_min)；

In S12: defining a closed area E in the matrix D, enabling elements in the area E to meet the sum of the maximum value and the sum of the maximum values of 32, recording coordinates of all elements in the area E, calculating the maximum row number difference and the maximum column number difference between any elements in the area E, and respectively recording the maximum row number difference and the maximum column number difference as N_LAnd N_W。

3. The infrared thermography detection method for the micro defects in the blades of the large wind turbine according to claim 1, wherein the infrared thermography detection method comprises the following steps: in S15, let ρ be the blade material density, c be the heat capacity, and K_pIs the x-y in-plane thermal conductivity, K, of the blade material_zFor the thermal conductivity of the blade material in the z direction (thickness direction), let

α＝K/(ρ·c)，α_z＝K_z/(ρ·c)，

The length of the blade is h₁Width of h₂Coefficient of K_X1＝2h₁l₁-X-ll₁-2nh₁l₁+2nll₁，K_X2＝X-ll₁+2nh₁l₁+2mll₁，K_Y1＝2h₂l₁-Y-wl₁-2mh₂l₁+2mwl₁，K_Y2＝Y-bl₁+2mh₂l₁+2mwl₁，K_Z1＝Z+dl₂+2pdl₂，K_Z2＝-Z+dl₂+2pdl₂；

Defining x-direction contrast

And contrast in the y-direction

And also the contrast in the thickness direction

In the formula, m, n, r and p are summation parameters, and the polynomial C is substituted by taking d as 0.0,0.1,0.2_x×C_y×C_zThe value of d, where the polynomial value is closest to 0.05, is the depth of defect value.

Images