A method and system for detecting defects inside a can body

Abstract

This invention relates to the field of defect detection technology, specifically to a method and system for detecting internal defects in tanks. The invention precisely quantifies the deviation at each wavelength in the detection spectrum by analyzing the difference between the detection spectrum and the baseline spectrum; further, it deeply analyzes the group dispersion of the detection spectrum by examining the differences between the detection spectra of different detection points; then, it meticulously evaluates the degree of anomaly in each band of the detection spectrum based on the differences between any band and other bands in the detection spectrum; and it fully considers the degree of abnormal overlap of band pairs by combining the correlation of overlapping band pairs in the detection spectra of two adjacent detection points; finally, it identifies several target wavelength bands, analyzes their longitudinal performance characteristics within the target wavelength bands, and examines their sensitivity, thereby enabling rapid and accurate location of internal defects in tanks, significantly improving detection efficiency and reliability.

Description

Technical Field

[0001] This invention relates to the field of defect detection technology, specifically to a method and system for detecting internal defects in tanks. Background Technology

[0002] In industrial production, tanks serve as critical containers for storing and transporting liquids, gases, and other media. Their structural integrity and safety directly impact product quality and public safety. Potential issues such as corrosion on the inner wall of the tank or media residue can lead to abnormalities in the internal material composition. These hidden dangers are easily overlooked by traditional visual inspections or destructive sampling tests, posing a threat to the tank's sealing and durability.

[0003] Against this backdrop, spectral analysis-based detection technology can acquire the spectral curves of several detection points in the internal region of the tank in real time—that is, the continuous response characteristics of the material to different wavelengths of light, including reflectivity, absorptivity, and transmittance—in a non-contact and high-precision manner, thereby achieving full-coverage screening and quantitative assessment of minute defects.

[0004] Spectroscopic techniques analyze the absorption, reflection, and emission characteristics of materials at different wavelengths of light to identify internal defects or anomalies. Current techniques rely on specific wavelengths corresponding to known defects to reflect different defect conditions within the container. However, the detection spectrum contains overall spectral characteristics and is influenced by various factors. Directly analyzing spectral curves at fixed detection wavelengths, ignoring the correlation between captured wavelengths, makes accurate defect analysis difficult. Furthermore, in the detection spectrum of a detection point, the detection areas of two points may overlap, resulting in the same region appearing in their spectral data, thus interfering with the identification of the corresponding wavelength range. Summary of the Invention

[0005] In order to solve the above technical problems, the purpose of this invention is to provide a method and system for detecting internal defects in tanks.

[0006] According to a first aspect of the present invention, a method for detecting internal defects in a tank is provided, the specific technical solution of which is as follows:

[0007] Collect the detection spectrum at the detection point on the tank, as well as the baseline spectrum of the defect-free area;

[0008] Analyze the difference between the detected spectrum and the baseline spectrum to obtain the deviation at each wavelength in the detected spectrum;

[0009] By analyzing the differences between the detection spectra of the detection point and other detection points, the group dispersion of the detection spectra of the detection point is obtained;

[0010] By analyzing the differences between any band in the detection spectrum and other bands, the degree of abnormality in each band of the detection spectrum can be obtained.

[0011] The correlation of overlapping band pairs in the detection spectra of two adjacent detection points is analyzed, and the degree of abnormal overlap of the band pairs is obtained by combining the degree of abnormality of the bands.

[0012] The target wavelength range is obtained by analyzing the band intersection range of the detection spectra of any two detection points.

[0013] By combining the deviation, the group dispersion, the degree of anomaly, and the degree of anomaly overlap, the sensitivity of the target wavelength band is obtained, and defect detection is completed.

[0014] In some embodiments of the present invention, analyzing the difference between the detection spectrum and the baseline spectrum to obtain the deviation at each wavelength of the detection spectrum includes:

[0015] For each wavelength, the reflectance difference between the detected spectrum and the baseline spectrum is analyzed, as well as the difference between the reflectance of the detected spectrum at each wavelength and the mean reflectance of the detected spectrum. Combined with the reflectance stability of the baseline spectrum, the deviation at each wavelength of the detected spectrum is obtained.

[0016] In some embodiments of the present invention, the differences between the detection spectra of the detection point and other detection points are analyzed to obtain the group dispersion of the detection spectra of the detection point, including:

[0017] For each detection point, the difference between the reflectance of the detection spectrum at each wavelength and the mean reflectance of the detection spectrum at the corresponding wavelength of all other detection points is analyzed. Combined with the reflectance stability of the detection spectra at the corresponding wavelength of all detection points, the population dispersion of the detection spectrum of the detection point is obtained.

[0018] In some embodiments of the present invention, analyzing the differences between any band and other bands in the detection spectrum to obtain the degree of abnormality of each band in the detection spectrum includes:

[0019] In the detected spectrum, a peak detection algorithm is used to locate all bands;

[0020] The dominance of the band is obtained based on the band width and band curvature in the detected spectrum;

[0021] For each detection point, the degree of difference between the reflectance of the detection spectrum at each band extreme point and the average reflectance of the detection spectrum at all band extreme points is analyzed. Combined with the dominance, the degree of anomaly in each band of the detection spectrum is obtained.

[0022] In some embodiments of the present invention, the dominance of a band is obtained based on the band width and band curvature in the detected spectrum, including:

[0023] Obtain the curvature at each wavelength corresponding to the band in the detection spectrum to obtain the curvature sequence of the band;

[0024] The mean curvature of the curvature sequence of the band is calculated, and the mean curvature difference between all two adjacent data points in the band is calculated to obtain the degree of slowness of the band's change trend.

[0025] The bandwidth is obtained based on the wavelength range corresponding to the band.

[0026] The dominance of the band is determined by combining the degree of slowness of the change trend with the band width.

[0027] In some embodiments of the present invention, the correlation of overlapping band pairs in the detection spectra of two adjacent detection points is analyzed, and the degree of abnormal overlap of the band pairs is obtained by combining the degree of anomaly of the bands, including:

[0028] In the detection spectra of any two adjacent detection points, the wavelength ranges corresponding to all their bands are obtained, and several band pairs with overlapping wavelength ranges are obtained.

[0029] The wavelength overlap range of the band pair is obtained, and the wavelength overlap degree of the band pair is obtained by combining the larger value of the wavelength range corresponding to the two bands in the band pair.

[0030] Calculate the Pearson correlation coefficient between the two bands in the band pair, and combine it with the degree of wavelength overlap to obtain the correlation of the band pair;

[0031] Based on the correlation, and combined with the average of the anomalies of the two bands in the band pair, the degree of abnormal overlap of the band pair is obtained.

[0032] In some embodiments of the present invention, the target wavelength band is obtained by analyzing the band intersection range of the detection spectra of any two detection points, including:

[0033] In all bands of the detection spectrum at all detection points, when two bands intersect and the intersection is greater than half the wavelength range of either band, the wavelength ranges corresponding to the two bands are merged to obtain several target wavelength bands.

[0034] In some embodiments of the present invention, the sensitivity of the target wavelength band is obtained by combining the deviation, the group dispersion, the degree of anomaly, and the degree of anomaly overlap, and defect detection of the tank is performed, including:

[0035] Based on the standard deviation of the deviation corresponding to each wavelength within the target wavelength range, and combined with the standard deviation of the band width corresponding to all bands within the target wavelength range, the longitudinal anisotropy of the bands within each target wavelength range is obtained.

[0036] The sensitivity coefficient of each detection point is obtained based on the population dispersion corresponding to the detection point, the degree of anomaly of the bands within the target wavelength range of the detection point, and the degree of abnormal overlap of the band pairs within the target wavelength range of the detection point.

[0037] By traversing all detection points within the target wavelength band, the sensitivity coefficient is calculated, and the sensitivity of the target wavelength band is obtained by combining the longitudinal anisotropy of the band.

[0038] Based on the sensitivity and the reflectivity of the target wavelength band, defects in the tank are detected.

[0039] According to a second aspect of the present invention, a tank internal defect detection system is provided, comprising: a memory and a processor, wherein:

[0040] The memory is used to store program code;

[0041] The processor is configured to read program code stored in the memory and execute the method described in the first aspect of the present invention.

[0042] In some embodiments of the present invention, the processor includes:

[0043] The spectral data acquisition module is used to acquire the detection spectrum at the detection point on the tank, as well as the baseline spectrum of the defect-free area;

[0044] The deviation analysis module is used to analyze the difference between the detected spectrum and the baseline spectrum to obtain the deviation at each wavelength of the detected spectrum;

[0045] The spectral population dispersion analysis module is used to analyze the differences between the detection spectra of the detection point and other detection points to obtain the population dispersion of the detection spectra of the detection point.

[0046] The band anomaly analysis module is used to analyze the differences between any band in the detection spectrum and other bands, and to obtain the anomaly degree of each band in the detection spectrum.

[0047] The band pair overlap analysis module is used to analyze the correlation of band pairs that have an intersection in the detection spectra of two adjacent detection points, and to obtain the degree of overlap of the band pairs by combining the degree of anomalousness of the bands.

[0048] The target wavelength band identification module is used to analyze the band intersection range of the detection spectra of any two detection points to obtain the target wavelength band.

[0049] The target wavelength band sensitivity analysis module is used to combine the deviation, the group dispersion, the degree of anomaly, and the degree of anomaly overlap to obtain the sensitivity of the target wavelength band and complete the defect detection.

[0050] Compared with existing technologies, the present invention provides a method and system for detecting internal defects in tanks, which has the following advantages:

[0051] This invention first establishes a baseline spectrum of defect-free samples to accurately quantify the deviation of the detection spectral wavelength, providing a benchmark for anomaly detection. Then, due to the complex redundancy of spectral features in the detection spectrum, the differences between the detection spectra of the specified detection point and those of other detection points are analyzed to deeply understand the group dispersion of the detection spectrum at each detection point and uncover data distribution characteristics. Based on this, the differences between any band in the detection spectrum and other bands are used to meticulously assess the degree of anomaly in a single spectral band, capturing subtle local changes. Furthermore, the correlation of overlapping band pairs in the detection spectra of two adjacent detection points is considered, fully taking into account minor overlaps between detection points to ensure data integrity and analytical accuracy. Finally, several target wavelength bands are successfully identified. By combining the longitudinal performance characteristics within these target wavelength bands and analyzing their sensitivity, internal defects in tanks can be quickly and accurately located, significantly improving detection efficiency and reliability, and effectively avoiding safety hazards and economic losses caused by missed or false detections in traditional detection methods. Attached Figure Description

[0052] To more clearly illustrate the technical solutions and advantages 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.

[0053] Figure 1 This is a schematic diagram of the basic process of a method for detecting internal defects in a tank according to an embodiment of the present invention.

[0054] Figure 2 A schematic diagram of the baseline spectral curve of a defect-free, normal, and qualified region provided in an embodiment of the present invention;

[0055] Figure 3 This is a schematic diagram of the detection spectrum curve of a detection point provided in one embodiment of the present invention;

[0056] Figure 4This is a schematic diagram of the detection spectral curves of two adjacent detection points provided in one embodiment of the present invention;

[0057] Figure 5 This is a schematic diagram of the basic components of a tank internal defect detection system provided in one embodiment of the present invention. Detailed Implementation

[0058] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a tank internal defect detection method and system proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. Terms such as “comprising,” “including,” or any other variations thereof are intended to cover a non-exclusive inclusion, such that a circuit structure, article, or device comprising a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such article or device. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of additional identical elements in the article or device that includes the element.

[0060] The specific scheme of the method for detecting internal defects of a tank provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0061] Please see Figure 1 This illustrates the basic process of a method for detecting internal defects in a tank provided by an embodiment of the present invention.

[0062] like Figure 1 As shown, an embodiment of the present invention provides a method for detecting internal defects in a tank, specifically including:

[0063] S100: Collects the detection spectrum at the tank inspection point, as well as the baseline spectrum of the defect-free area.

[0064] Several detection points were selected inside the tank, and the arc spectrum was collected using a fiber optic probe or a linear CCD spectrometer, covering the ultraviolet-visible-near-infrared band (200-1100nm). A diffuse reflection light source (such as a bowl-shaped LED) was used to suppress interference from the tank's high reflectivity. Reflectance detection spectral curves were obtained for each detection point, where the x-axis represents the spectral wavelength in nm, and the y-axis represents the reflectance in %. Figure 2 As shown.

[0065] In addition, baseline spectral curves of defect-free, normal, and acceptable areas are collected as a reference for the normal state, such as... Figure 2 As shown.

[0066] S200: Analyze the difference between the detection spectrum and the baseline spectrum to obtain the deviation at each wavelength on the detection spectrum.

[0067] Reflectivity reflects the ability of a material surface to reflect incident light. Defects (such as cracks and oxidation) can change the surface morphology and composition, leading to abnormal reflectivity spectra. Therefore, when the detection point is located in a defect segment, there will be a large difference between the reflectivity values ​​of the detection spectrum and the baseline spectrum at a certain wavelength.

[0068] Based on the above analysis, in the embodiments of the present invention, the deviation at each wavelength of the detection spectrum is obtained by analyzing the difference between the detection spectrum and the baseline spectrum. Specifically, for each wavelength, the difference in reflectance between the detection spectrum and the baseline spectrum is analyzed, as well as the difference between the reflectance of the detection spectrum at each wavelength and the mean reflectance of the detection spectrum. Combined with the reflectance stability of the baseline spectrum, the deviation at each wavelength of the detection spectrum is obtained. The formula for calculating the deviation at wavelength x of the detection spectrum is constructed as follows:

[0069]

[0070] In the formula, F x Indicates the deviation at wavelength x in the detection spectrum; f x f represents the reflectance value of the detected spectrum at wavelength x; 0x Δf represents the reflectance value of the baseline spectrum at wavelength x; x σ(f0) represents the difference between the reflectance value of the detected spectrum at wavelength x and the mean reflectance of the detected spectrum at all wavelengths; σ(f0) represents the standard deviation of the reflectance of the baseline spectrum at all wavelengths; ε represents a constant, which can be 0.1 to prevent the denominator from being 0.

[0071] In the detection spectrum, when the reflectance values ​​of the detection spectrum and the baseline spectrum at wavelength x differ (|f x -f 0x The larger the |), the greater the difference (Δf) between the reflectance value of the detected spectrum at wavelength x and the average reflectance value of the detected spectrum at all wavelengths. x The larger the standard deviation (σ(f0)) is, the greater the deviation between the detected spectrum and the baseline spectrum at wavelength x; and the more stable the reflectance distribution of the baseline spectrum, that is, the smaller the standard deviation (σ(f0)) of the baseline spectrum, the greater the deviation of the abnormal distribution of the reflectance value of the detected spectrum at wavelength x.

[0072] S300: Analyze the differences between the detection spectra of the detection point and other detection points to obtain the group dispersion of the detection spectra of the detection point.

[0073] Among all the selected detection points, the more significant the abnormal difference between the spectral data of a detection point and the spectral data of the other detection points, the more likely that the location of the detection point is within the abnormal range inside the tank.

[0074] Therefore, in the embodiments of the present invention, the population dispersion of the detection spectrum of the detection point is obtained by analyzing the differences between the detection spectra of the detection point and those of other detection points. Specifically, the population dispersion of the detection spectrum of each detection point can be determined based on the differences in reflectance values ​​at different wavelengths between the spectral curves. Therefore, for each detection point, the difference between the reflectance of the detection spectrum of the detection point at each wavelength and the average reflectance of the detection spectra of all other detection points at the corresponding wavelength is analyzed. Combined with the reflectance stability of the detection spectra of all detection points at the corresponding wavelength, the population dispersion of the detection spectrum of the detection point is obtained. The formula for calculating the population dispersion of the detection spectrum of the detection point is as follows:

[0075]

[0076] In the formula, L a f represents the population dispersion of the detection spectrum at detection point a; a (λ x f(λ) represents the reflectance value at wavelength x in the detection spectrum of detection point a; x ) represents the mean reflectance value of all detection points at wavelength x; σ f (λ x ) represents the standard deviation of reflectance at wavelength x for all detection points; n represents the number of wavelengths in the detection spectrum of detection point a; ε represents a constant, which can be 0.1 to prevent the denominator from being 0.

[0077] When f a (λ x )-f(λ x The larger the σ value, that is, the greater the difference between the reflectance value at each wavelength and the overall reflectance value at that wavelength in the detection spectrum at detection point a, the stronger the group dispersion of the detection spectrum at that detection point; and when σ f (λ x The larger the value, that is, the larger the standard deviation of the reflectance values ​​of all detection points at a certain wavelength, the more unstable the reflectance value at that wavelength and the stronger the group dispersion.

[0078] S400: Analyzes the differences between any band in the detection spectrum and other bands to obtain the degree of abnormality of each band in the detection spectrum.

[0079] At any given detection point, the spectrometer emits light and detects and recovers it. The detection area may contain qualified tank sections or tank defects. Therefore, the spectral data from a single monitoring point should include several bands, each representing a different wavelength range of reflection within that detection area—that is, different content regions. A significant deviation from the overall reflectance range with a sudden increase or decrease in reflectance within a specific wavelength range indicates an abnormal band signal.

[0080] In the detection spectrum at a given testing point, the main body of the tank should correspond to a wider range of wavelengths in the spectral data, and these wavelengths should change gradually. Therefore, for any given wavelength at a testing point, we need to analyze whether the corresponding wavelength range of each wavelength is more likely to be a qualified area within the tank. Figure 3 As shown, the A band is more dominant and is more likely to be the wavelength band corresponding to the qualified area of ​​the tank.

[0081] Based on the above analysis, in the embodiments of the present invention, the degree of abnormality of each band in the detection spectrum is obtained by analyzing the differences between any band and other bands in the detection spectrum. Further aspects include:

[0082] First, in the detection spectrum of the detection point, the peak detection algorithm is used to locate all bands.

[0083] Then, the dominance of a band is obtained based on the band width and band curvature in the detection spectrum. Specifically, the curvature at each wavelength corresponding to the band in the detection spectrum is obtained to obtain the curvature sequence of the band; the mean curvature of the band curvature sequence is calculated, and the mean curvature difference between all two adjacent data points in the band is calculated to obtain the degree of slowness of the band's change trend; the band width is obtained based on the wavelength range corresponding to the band; and the dominance of the band is obtained by combining the degree of slowness of the change trend with the band width. The formula for calculating the dominance of band y in the detection spectrum is constructed as follows:

[0084]

[0085] In the formula, D(λ′) y w(λ′) represents the dominance of band y in the detection spectrum; y ) represents the bandwidth of band y in the detection spectrum; k(λ′) y ) represents the mean of all curvatures in the curvature sequence corresponding to band y in the detection spectrum; Δk(λ′) y ) represents the mean of the difference between two adjacent curvature values ​​(the absolute value of the difference between two adjacent curvature values) in the curvature sequence corresponding to band y in the detection spectrum.

[0086] When the width of band y is wider, i.e. w(λ′) yThe larger the value of k(λ′), the larger the wavelength range corresponding to band y, and the stronger its dominance. Furthermore, since the detection spectral data corresponding to a dominant qualified tank area should exhibit slowly changing bands, the smaller the curvature of all data points in band y of the detection spectrum, i.e., the smaller k(λ′) value, the stronger the dominance. y The smaller the value of Δk(λ′), and the smaller the curvature difference between two adjacent data points, i.e., Δk(λ′) y The smaller the value, the stronger the dominance of its corresponding band.

[0087] Finally, for all bands in the detection spectrum at a detection point, the stronger the dominance of a band, meaning its corresponding wavelength range is more likely to represent the spectrum within the acceptable range for that detection point, the weaker the dominance, the greater the degree of anomaly in a single spectrum. Furthermore, in the detection spectrum at a detection point, the smaller the reflectance value of a band, and the more prominent the reflectance value of a particular band in the spectrum at that detection point, the higher the degree of anomaly. Therefore, for each detection point, the difference between the reflectance at each band's extreme point and the average reflectance at all band extreme points is analyzed. Combined with dominance, the degree of anomaly in each band of the detection spectrum is obtained. The formula for calculating the degree of anomaly of band y in the detection spectrum of detection point a is constructed as follows:

[0088]

[0089] In the formula, Y a (λ′ y ) represents the degree of anomaly in band y in the detection spectrum of detection point a; f a (λ′ y ) represents the reflectance value of the detection spectrum at the extreme point of band y of detection point a; D(λ′) represents the mean reflectance at all extreme points in the detection spectrum of detection point a; y ) indicates the dominance of band y in the detection spectrum; ε represents a constant, which can be 0.1 to prevent the denominator from being 0.

[0090] The smaller the reflectance value at the extreme point of a band, and the more the overall reflectance value at all extreme points of bands in the detection spectrum at that detection point deviates, the higher the degree of anomaly; the lower the dominance of the wavelength range corresponding to the band, the higher the degree of anomaly of that band.

[0091] This yields the degree of anomaly in all bands of the detection spectrum at all detection points, and obtains the wavelength range of the corresponding bands.

[0092] S500: Analyze the correlation of band pairs that have intersection in the detection spectra of two adjacent detection points, and combine the degree of band anomaly to obtain the degree of abnormal overlap of the band pairs.

[0093] When performing spectral analysis on the inside of the tank, the monitoring areas of the detection points may overlap. Considering the possibility of overlap between adjacent monitoring points, when overlap occurs, the overlapping band pairs will show a strong correlation, and the wavelength overlap range will be larger.

[0094] Based on the above analysis, in the embodiments of the present invention, the degree of abnormal overlap of band pairs is obtained by analyzing the correlation of overlapping band pairs in the detection spectra of two adjacent detection points and combining the degree of band anomaly. Specifically, in the detection spectra of any two adjacent detection points, the wavelength ranges corresponding to all their bands are obtained, and several band pairs with overlapping wavelength ranges are obtained (the two bands in a band pair come from two adjacent detection points respectively), such as... Figure 4 As shown; obtain the wavelength overlap range of the band pair, and combine it with the larger value of the wavelength range corresponding to the two bands in the band pair to obtain the wavelength overlap degree of the band pair; calculate the Pearson correlation coefficient between the two bands in the band pair, and combine it with the wavelength overlap degree to obtain the correlation of the band pair; based on the correlation, and combined with the mean of the anomaly degree corresponding to the two bands in the band pair, obtain the abnormal overlap degree of the band pair. The formula for calculating the abnormal overlap degree of the band pair is as follows:

[0095]

[0096] In the formula, H represents the degree of abnormal overlap between band pairs; J a,b (λ′) represents the intersection of the wavelength ranges corresponding to the two bands a and b in the band pair; max(λ′) represents the larger value of the wavelength ranges of the two bands a and b in the band pair; ρ a,b (λ′) represents the Pearson correlation coefficient between two bands a and b in the band pair; This represents the average degree of anomalousness between the two bands a and b in the band pair.

[0097] In the detection spectra of two adjacent detection points, when the relative wavelength ranges of the two bands intersect... The larger the value, the greater the Pearson correlation coefficient between the two bands (ρ). a,b (λ′)), the stronger the correlation of the waveforms, the more likely they are to overlap; and at this time, it is necessary to exclude the overlap of the corresponding wavebands in the qualified area of ​​the tank. Therefore, when the average degree of abnormality of the two wavebands is aligned, The larger the value, the greater the abnormal overlap of the band pair.

[0098] S600: Analyze the band intersection range of the detection spectra of any two detection points to obtain the target wavelength band.

[0099] In all bands of the detection spectrum at all detection points, when two bands intersect and the intersection is greater than half the wavelength range of either band, the corresponding wavelength ranges of the two bands are merged, thus obtaining several target wavelength bands on the wavelength coordinate axis of the detection spectrum curve. It should be noted that the number of bands at the detection points that meet the merging condition may be two or more.

[0100] S700: By combining deviation, group dispersion, degree of anomaly, and degree of anomaly overlap, the sensitivity of the target wavelength band is obtained, and defect detection is completed.

[0101] The greater the degree of anomaly in the target wavelength band across all detection points, and the stronger the group dispersion of the detection spectrum at the corresponding detection point, the stronger the sensitivity of that detection point in the corresponding target wavelength band.

[0102] Therefore, in embodiments of the present invention, by combining deviation, group dispersion, anomaly degree, and anomaly overlap degree, the sensitivity of the target wavelength band is obtained, thus completing defect detection. Further aspects include:

[0103] First, for the target wavelength range on the wavelength coordinate axis of the detection spectrum, in order to determine the sensitive wavelength range, it is also necessary to analyze the longitudinal differences of each band within the target wavelength range. That is, for a band at a certain monitoring point within the target wavelength range, the greater the significant deviation of that band in the longitudinal direction, the more it promotes the sensitivity of that target wavelength range. Therefore, based on the standard deviation of the deviation corresponding to each wavelength within the target wavelength range, combined with the standard deviation of the bandwidth corresponding to all bands within the target wavelength range, the longitudinal anisotropy of the bands within each target wavelength range is obtained.

[0104]

[0105] In the formula, Z b σ(F) represents the longitudinal anisotropy of the bands within the target wavelength band b; σ[w(λ′)] represents the standard deviation of the deviation between the detected spectrum and the baseline spectrum of all detection points corresponding to each wavelength within the target wavelength band b; σ[w(λ′)] represents the standard deviation of the bandwidth of all bands within the target wavelength band b; ε represents a constant, which can be 0.1 to prevent the denominator from being 0.

[0106] The larger the standard deviation of the deviation, i.e., the larger the value of σ(Δf), the more unstable the reflectance values ​​between detection points within the target wavelength band are, and the greater the longitudinal anisotropy within the target wavelength band. The smaller the standard deviation of the bandwidth, i.e., the smaller the value of σ[w(λ′)], the stronger the reference value of the reflectance values ​​of the band in the detection spectrum. The larger the standard deviation of the bandwidth, the more different the meanings of the content reflected inside the tank may be, and therefore the less reference value of the spectral deviation is, and the smaller the longitudinal anisotropy of the band within the target wavelength band is.

[0107] Then, based on the population dispersion corresponding to the detection point, the degree of anomaly of the bands within the target wavelength band of the detection point, and the degree of abnormal overlap of band pairs within the target wavelength band of the detection point, the sensitivity coefficient of each detection point is obtained; and by traversing all detection points included in the target wavelength band, calculating the sensitivity coefficient, and combining it with the longitudinal anisotropy of the bands, the sensitivity of the target wavelength band is obtained. The formula for calculating the sensitivity of the target wavelength band b is as follows:

[0108]

[0109] In the formula, M b Z represents the sensitivity of the target wavelength band b; b Indicates the longitudinal anisotropy of the band within the target wavelength band b; L i,b Y represents the population dispersion of the detection spectrum of detection point i within the target wavelength band b; i,b (λ′ y (H) indicates the degree of anomaly in the target wavelength band b at detection point i; i,b ) max N represents the maximum value of the abnormal overlap of the corresponding band pair within the target wavelength band b at detection point i; b This indicates the number of detection points contained within the target wavelength band b.

[0110] For a target wavelength band, the greater the degree of anomaly of the detection point within that target wavelength band, and the greater the overall group dispersion of the detection spectrum of that detection point, the more significant the anomaly of that band in the target wavelength band, and the positive feedback exists in the quantification of its sensitivity to that target wavelength band. However, when there is anomaly overlap of bands, there is negative feedback in the quantification of its sensitivity to the target wavelength band. Therefore, the greater the anomaly overlap, the lower the sensitivity of the target wavelength band.

[0111] This yields several target wavelength bands within the entire detection spectrum, as well as the sensitivity of each target wavelength band. The higher the sensitivity of a target wavelength band, the stronger the correlation between the spectral characteristics within that band and the tank defect, and the more likely it is to correspond to the characteristic wavelength range of the defect in the detection spectrum.

[0112] Among all target wavelength bands, they are sorted in descending order of sensitivity, forming an ordered sensitivity sequence. From all data, the two sensitivity data points corresponding to the maximum difference between two adjacent sensitivity data points are identified. Based on these two sensitivity data points and their left and right sides, the ordered sensitivity sequence is divided into two parts: the part with the larger value is recorded as the sensitive data, and the target wavelength band corresponding to the sensitive data is recorded as the sensitive wavelength band. This yields all sensitive wavelength bands and their sensitivities. Therefore, the sensitive wavelength band with higher sensitivity corresponds to the characteristic band range of the defect in the detection spectrum. The sensitive wavelength band may correspond to a defect in the tank caused by material changes due to certain factors; obtaining the location of this sensitive wavelength band indicates the possible location of the defect.

[0113] Finally, when performing defect detection inside the tank, the tank is inspected based on the sensitivity and reflectivity of the target wavelength band. Specifically, the possible locations of defects are confirmed by the sensitivity and reflectivity of the sensitive wavelength band, thus achieving accurate identification and location of defects inside the tank.

[0114] Please see Figure 5 This illustrates the basic components of a tank internal defect detection system provided by an embodiment of the present invention.

[0115] like Figure 5 As shown, a tank internal defect detection system includes: a memory 10 and a processor 20, wherein:

[0116] Memory 10 is used to store program code;

[0117] The processor 20 is used to read the program code stored in the memory 10 and execute the following functions: acquire the detection spectrum at the detection point on the tank and the baseline spectrum of the defect-free area; analyze the difference between the detection spectrum and the baseline spectrum to obtain the deviation at each wavelength on the detection spectrum; analyze the difference between the detection spectra of the detection point and other detection points to obtain the group dispersion of the detection spectrum of the detection point; analyze the difference between any band in the detection spectrum and other bands to obtain the degree of abnormality of each band in the detection spectrum; analyze the correlation of band pairs with intersection in the detection spectra of two adjacent detection points, and combine the degree of abnormality of the bands to obtain the degree of abnormal overlap of the band pairs; analyze the band intersection range of the detection spectra of any two detection points to obtain the target wavelength band; and combine the deviation, group dispersion, degree of abnormality, and degree of abnormal overlap to obtain the sensitivity of the target wavelength band, thus completing the defect detection.

[0118] Furthermore, the processor 20 includes a spectral data acquisition module 21, a deviation analysis module 22, a spectral population dispersion analysis module 23, a band anomaly analysis module 24, a band pair anomaly overlap analysis module 25, a target wavelength band identification module 26, and a target wavelength band sensitivity analysis module 27. Wherein:

[0119] The spectral data acquisition module 21 is used to acquire the detection spectrum at the detection point of the tank and the baseline spectrum of the defect-free area;

[0120] The deviation analysis module 22 is used to analyze the difference between the detection spectrum and the baseline spectrum to obtain the deviation at each wavelength on the detection spectrum;

[0121] The spectral population dispersion analysis module 23 is used to analyze the differences between the detection spectra of the detection point and other detection points to obtain the population dispersion of the detection spectra of the detection point.

[0122] The band anomaly analysis module 24 is used to analyze the differences between any band in the detection spectrum and other bands, and to obtain the anomaly degree of each band in the detection spectrum.

[0123] The band pair anomalous overlap degree analysis module 25 is used to analyze the correlation of band pairs that have an intersection in the detection spectra of two adjacent detection points, and obtain the anomalous overlap degree of the band pairs by combining the anomalous degree of the bands.

[0124] The target wavelength band identification module 26 is used to analyze the band intersection range of the detection spectrum of any two detection points to obtain the target wavelength band;

[0125] The target wavelength band sensitivity analysis module 27 is used to combine deviation, group dispersion, degree of anomaly and degree of anomaly overlap to obtain the sensitivity of the target wavelength band and complete defect detection.

[0126] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0127] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for detecting internal defects in a tank, characterized in that, The method includes: Collect the detection spectrum at the tank inspection points, as well as the baseline spectrum of the defect-free area; collect the detection spectrum at the tank inspection points, including: collecting the arc spectrum using a fiber optic probe or linear CCD spectrometer, covering the ultraviolet-visible-near-infrared band; obtain the reflectance detection spectrum curve for each inspection point; The difference between the detected spectrum and the baseline spectrum is analyzed to obtain the deviation at each wavelength of the detected spectrum. This includes: analyzing the difference in reflectance between the detected spectrum and the baseline spectrum for each wavelength, and analyzing the difference between the reflectance of the detected spectrum at each wavelength and the mean reflectance of the detected spectrum. Combined with the reflectance stability of the baseline spectrum, the deviation at each wavelength of the detected spectrum is obtained. The calculation formula is: In the formula, Indicates the wavelength in the detection spectrum Deviation at; Indicates the detection spectrum at wavelength The reflectance value at that location; Indicates the baseline spectrum at wavelength The reflectance value at that location; Indicates the detection spectrum at wavelength The difference between the reflectance value at a given point and the mean reflectance of the detected spectrum at all wavelengths; This represents the standard deviation of reflectance of the baseline spectrum at all wavelengths; This represents a constant with a value of 0.

1. The differences between the detection spectra of the detection point and those of other detection points are analyzed to obtain the group dispersion of the detection spectra of the detection point. This includes: for each detection point, analyzing the difference between the reflectance of the detection spectrum of the detection point at each wavelength and the mean reflectance of the detection spectra of all other detection points at the corresponding wavelength; combining this with the reflectance stability of the detection spectra of all detection points at the corresponding wavelength, the group dispersion of the detection spectrum of the detection point is obtained. The calculation formula is as follows: In the formula, Indicates the detection point The population dispersion of the detection spectrum; Indicates the detection point In the detection spectrum at wavelength The reflectance value at that location; This indicates that all detection points are at wavelength The average reflectance value at that location; This indicates that all detection points are at wavelength The standard deviation of reflectance at that location; Indicates the detection point The number of all wavelengths in the detection spectrum; Analyze the differences between any band and other bands in the detection spectrum to obtain the degree of anomaly in each band of the detection spectrum, including: using a peak detection algorithm to locate all bands in the detection spectrum; The dominance of a band is determined based on the band width and band curvature in the detection spectrum, including: obtaining the curvature at each wavelength corresponding to the band in the detection spectrum, and obtaining the curvature sequence of the band; The mean curvature of the curvature sequence of the band is calculated, and the mean curvature difference between all two adjacent data points in the band is calculated to obtain the degree of slowness of the band's change trend. The bandwidth is obtained based on the wavelength range corresponding to the band. Combining the degree of slowness of the change trend with the band width, the dominance of the band is obtained, calculated using the following formula: In the formula, Indicates the band in the detection spectrum Dominance; Indicates the band in the detection spectrum The bandwidth; Indicates the band in the detection spectrum The mean of all curvatures in the corresponding curvature sequence; Indicates the band in the detection spectrum The mean difference between two adjacent curvature values ​​in the corresponding curvature sequence; For each detection point, the difference between the reflectance at each band's extreme point and the mean reflectance at all band extreme points in the detection spectrum is analyzed. Combined with dominance, the degree of anomaly in each band of the detection spectrum is obtained, calculated using the following formula: In the formula, Indicates the detection point The detection spectrum in the mid-band The degree of abnormality; Indicates the detection point The detection spectrum in the band Reflectance value at extreme points; Indicates the detection point The mean reflectance at all extreme points in the detected spectrum; Indicates the band in the detection spectrum Dominance; The correlation of band pairs that overlap in the detection spectra of two adjacent detection points is analyzed. Combined with the degree of anomalousness of the bands in the band pairs, the degree of anomalous overlap of the band pairs is obtained. This includes: obtaining the wavelength range corresponding to all bands in the detection spectra of any two adjacent detection points, and obtaining several band pairs whose wavelength ranges overlap. The wavelength overlap range of the band pair is obtained, and the larger value of the wavelength range corresponding to the two bands in the band pair is combined to obtain the wavelength overlap degree of the band pair. Calculate the Pearson correlation coefficient between the two bands in a band pair, and combine it with the degree of wavelength overlap to obtain the correlation of the band pair; Based on the correlation and the average anomaly severity of the two bands in the band pair, the degree of overlap of the band pair is obtained. The calculation formula is as follows: In the formula, Indicates the degree of abnormal overlap between band pairs; Indicates the two bands in the band pair. and The intersection of the corresponding wavelength ranges; Indicates the two bands in the band pair. and The larger value in the wavelength range; Indicates the two bands in the band pair. and The Pearson correlation coefficient; Indicates the two bands in the band pair. and The mean of the degree of abnormality; Analyze the band intersection range of the bands in the detection spectrum of any two detection points to obtain the target wavelength range, including: when two bands have an intersection in all bands of the detection spectrum of all detection points and the intersection is greater than half of the wavelength range of either band, merge the wavelength ranges corresponding to the two bands to obtain several target wavelength ranges. By combining deviation, group dispersion, anomaly degree, and anomaly overlap degree, the sensitivity of the target wavelength band is obtained, and defect detection is completed, including: Based on the standard deviation of the deviation corresponding to each wavelength within the target wavelength band, and combined with the standard deviation of the bandwidth corresponding to all bands within the target wavelength band, the longitudinal anisotropy of the bands within each target wavelength band is obtained. The calculation formula is as follows: In the formula, Indicates the target wavelength band Longitudinal anisotropy of bands within the band; Indicates the target wavelength range The standard deviation of the deviation between the detected spectrum and the baseline spectrum at all detection points corresponding to each wavelength; Indicates the target wavelength range The standard deviation of the bandwidth across all bands within the range; The sensitivity coefficient of each detection point is obtained based on the population dispersion corresponding to the detection point, the degree of anomaly of the bands within the target wavelength range of the detection point, and the degree of abnormal overlap of the band pairs within the target wavelength range of the detection point. By traversing all detection points within the target wavelength band, calculating the sensitivity coefficient, and combining this with the longitudinal anisotropy of the band, the sensitivity of the target wavelength band is obtained. The calculation formula is as follows: In the formula, Indicates the target wavelength band Sensitivity; Indicates the target wavelength band Longitudinal anisotropy of bands within the band; Indicates the target wavelength band Included detection points The population dispersion of the detection spectrum; Indicates at the testing point Target wavelength band The degree of anomaly in the corresponding band; Indicates at the testing point Target wavelength band The maximum value of the abnormal overlap of corresponding band pairs within the range; Indicates the target wavelength band The number of detection points contained therein; Based on sensitivity and the reflectivity of the target wavelength band, defects in the tank are detected.

2. A tank internal defect detection system, characterized in that, The system includes: a memory and a processor, wherein: The memory is used to store program code; The processor is configured to read the program code stored in the memory and execute the method as described in claim 1.

3. The tank internal defect detection system according to claim 2, characterized in that, The processor includes: The spectral data acquisition module is used to acquire the detection spectrum at the detection point on the tank, as well as the baseline spectrum of the defect-free area; The deviation analysis module is used to analyze the difference between the detected spectrum and the baseline spectrum to obtain the deviation at each wavelength of the detected spectrum; The deviation analysis module is specifically used to analyze the reflectance difference between the detected spectrum and the baseline spectrum for each wavelength, as well as the difference between the reflectance of the detected spectrum at each wavelength and the mean reflectance of the detected spectrum. Combined with the reflectance stability of the baseline spectrum, the deviation at each wavelength of the detected spectrum is obtained. The calculation formula is as follows: In the formula, Indicates the wavelength in the detection spectrum Deviation at; Indicates the detection spectrum at wavelength The reflectance value at that location; Indicates the baseline spectrum at wavelength The reflectance value at that location; Indicates the detection spectrum at wavelength The difference between the reflectance value at a given point and the mean reflectance of the detected spectrum at all wavelengths; This represents the standard deviation of reflectance of the baseline spectrum at all wavelengths; This represents a constant with a value of 0.

1. The spectral population dispersion analysis module is used to analyze the differences between the detection spectra of the detection point and other detection points to obtain the population dispersion of the detection spectra of the detection point. The spectral population dispersion analysis module is specifically used to analyze, for each detection point, the difference between the reflectance of the detection spectrum at each wavelength and the mean reflectance of the detection spectra at the corresponding wavelength of all other detection points. Combined with the reflectance stability of the detection spectra at the corresponding wavelengths of all detection points, the population dispersion of the detection spectrum at each detection point is obtained. The calculation formula is as follows: In the formula, Indicates the detection point The population dispersion of the detection spectrum; Indicates the detection point In the detection spectrum at wavelength The reflectance value at that location; This indicates that all detection points are at wavelength The average reflectance value at that location; This indicates that all detection points are at wavelength The standard deviation of reflectance at that location; Indicates the detection point The number of all wavelengths in the detection spectrum; The band anomaly analysis module is used to analyze the differences between any band in the detection spectrum and other bands, and to obtain the anomaly degree of each band in the detection spectrum. The band anomaly analysis module is specifically used to locate all bands in the detected spectrum using a peak detection algorithm; The dominance of a band is determined by the band width and band curvature in the detected spectrum. For each detection point, the difference between the reflectance at each band's extreme point and the mean reflectance at all band extreme points in the detection spectrum is analyzed. Combined with dominance, the degree of anomaly in each band of the detection spectrum is obtained, calculated using the following formula: In the formula, Indicates the detection point The detection spectrum in the mid-band The degree of abnormality; Indicates the detection point The detection spectrum in the band Reflectance value at extreme points; Indicates the detection point The mean reflectance at all extreme points in the detected spectrum; Indicates the band in the detection spectrum Dominance; The band pair overlap analysis module is used to analyze the correlation of band pairs that have an intersection in the detection spectra of two adjacent detection points, and to obtain the degree of overlap of the band pairs by combining the degree of anomalousness of the bands in the band pairs. The band anomaly analysis module is specifically used to obtain the curvature at each wavelength corresponding to the band in the detection spectrum, and to obtain the curvature sequence of the band. The mean curvature of the curvature sequence of the band is calculated, and the mean curvature difference between all two adjacent data points in the band is calculated to obtain the degree of slowness of the band's change trend. The bandwidth is obtained based on the wavelength range corresponding to the band. Combining the degree of slowness of the change trend with the band width, the dominance of the band is obtained, calculated using the following formula: In the formula, Indicates the band in the detection spectrum Dominance; Indicates the band in the detection spectrum The bandwidth; Indicates the band in the detection spectrum The mean of all curvatures in the corresponding curvature sequence; Indicates the band in the detection spectrum The mean difference between two adjacent curvature values ​​in the corresponding curvature sequence; The band pair overlap analysis module is specifically used to obtain the wavelength range corresponding to all bands in the detection spectrum of any two adjacent detection points, and to obtain several band pairs whose wavelength ranges overlap. The wavelength overlap range of the band pair is obtained, and the larger value of the wavelength range corresponding to the two bands in the band pair is combined to obtain the wavelength overlap degree of the band pair. Calculate the Pearson correlation coefficient between the two bands in a band pair, and combine it with the degree of wavelength overlap to obtain the correlation of the band pair; Based on the correlation and the average anomaly severity of the two bands in the band pair, the degree of overlap of the band pair is obtained. The calculation formula is as follows: In the formula, Indicates the degree of abnormal overlap between band pairs; Indicates the two bands in the band pair. and The intersection of the corresponding wavelength ranges; Indicates the two bands in the band pair. and The larger value in the wavelength range; Indicates the two bands in the band pair. and The Pearson correlation coefficient; Indicates the two bands in the band pair. and The mean of the degree of abnormality; The target wavelength band identification module is used to analyze the band intersection range of the detection spectra of any two detection points to obtain the target wavelength band. The target wavelength band identification module is specifically used to merge the wavelength ranges of two bands in all bands of the detection spectrum at all detection points, when there is an intersection between two bands and the intersection is greater than half of the wavelength range of either band, to obtain several target wavelength bands. The target wavelength band sensitivity analysis module is used to combine the deviation, the group dispersion, the degree of anomaly, and the degree of anomaly overlap to obtain the sensitivity of the target wavelength band and complete the defect detection. The target wavelength band sensitivity analysis module is specifically used to obtain the longitudinal anisotropy of each target wavelength band based on the standard deviation of the deviation corresponding to each wavelength within the target wavelength band, combined with the standard deviation of the bandwidth corresponding to all bands within the target wavelength band. The calculation formula is as follows: In the formula, Indicates the target wavelength band Longitudinal anisotropy of bands within the band; Indicates the target wavelength range The standard deviation of the deviation between the detected spectrum and the baseline spectrum at all detection points corresponding to each wavelength; Indicates the target wavelength range The standard deviation of the bandwidth across all bands within the range; The sensitivity coefficient of each detection point is obtained based on the population dispersion corresponding to the detection point, the degree of anomaly of the bands within the target wavelength range of the detection point, and the degree of abnormal overlap of the band pairs within the target wavelength range of the detection point. By traversing all detection points within the target wavelength band, the sensitivity coefficient is calculated. Combined with the longitudinal anisotropy of the band, the sensitivity of the target wavelength band is obtained. The calculation formula is as follows: In the formula, Indicates the target wavelength band Sensitivity; Indicates the target wavelength band Longitudinal anisotropy of bands within the band; Indicates the target wavelength band Included detection points The population dispersion of the detection spectrum; Indicates at the testing point Target wavelength band The degree of anomaly in the corresponding band; Indicates at the testing point Target wavelength band The maximum value of the abnormal overlap of corresponding band pairs within the range; Indicates the target wavelength band The number of detection points contained therein; Based on the sensitivity and the reflectivity of the target wavelength band, defects in the tank are detected.

Images