CN115575331B - In-situ measurement method for surface wettability distribution of insulating material based on spectrum inversion - Google Patents

In-situ measurement method for surface wettability distribution of insulating material based on spectrum inversion Download PDF

Info

Publication number
CN115575331B
CN115575331B CN202210478805.1A CN202210478805A CN115575331B CN 115575331 B CN115575331 B CN 115575331B CN 202210478805 A CN202210478805 A CN 202210478805A CN 115575331 B CN115575331 B CN 115575331B
Authority
CN
China
Prior art keywords
lambda
insulating material
data
wettability
surface wettability
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.)
Active
Application number
CN202210478805.1A
Other languages
Chinese (zh)
Other versions
CN115575331A (en
Inventor
夏昌杰
任明
李乾宇
董明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian Jiaotong University
Original Assignee
Xian Jiaotong University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Xian Jiaotong University filed Critical Xian Jiaotong University
Priority to CN202210478805.1A priority Critical patent/CN115575331B/en
Publication of CN115575331A publication Critical patent/CN115575331A/en
Application granted granted Critical
Publication of CN115575331B publication Critical patent/CN115575331B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N13/00—Investigating surface or boundary effects, e.g. wetting power; Investigating diffusion effects; Analysing materials by determining surface, boundary, or diffusion effects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
    • G06F17/13—Differential equations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Computational Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Immunology (AREA)
  • Biochemistry (AREA)
  • Analytical Chemistry (AREA)
  • Chemical & Material Sciences (AREA)
  • Operations Research (AREA)
  • Algebra (AREA)
  • Health & Medical Sciences (AREA)
  • Pathology (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)

Abstract

The method comprises the steps of obtaining surface wettability spectrum image data of an insulating material by utilizing a spectrum image technology, preprocessing the surface wettability spectrum image data of the insulating material, extracting regression analysis characteristic wave bands of the surface wettability spectrum image of the insulating material, establishing a regression analysis model of the surface wettability spectrum image of the insulating material, and realizing dynamic in-situ detection of the surface wettability of the insulating material under the operating voltage. By combining the spectrum information and the space image information, the method is suitable for detecting the surface wettability of different insulating materials, and dynamically detects the local drying zone caused by the thermal effect under the operation voltage, so that guidance is provided for the on-line detection of the power equipment and the formulation of the operation and maintenance strategy of the power equipment, and basis and reference are provided for the design and dampproof design of the insulating structure in different power equipment.

Description

In-situ measurement method for surface wettability distribution of insulating material based on spectrum inversion
Technical Field
The invention belongs to the technical field of online detection of surface wetting of an insulating material of power equipment, and particularly relates to an in-situ measurement method of surface wettability distribution of the insulating material based on spectrum inversion.
Background
In the field of high voltage and insulation, the insulation material plays a role of mechanical support and electrical insulation, however, as the insulation material works in the atmosphere for a long time, the surface of the insulation material inevitably suffers from the influence of factors such as water vapor and the like to cause the surface to be wet, so that the surface insulation performance is greatly reduced, and particularly, the phenomenon is more obvious after the surface of the insulation material accumulates dirt. Research shows that in the process of the surface flashover of the insulating material, partial filth is dissolved and the surface leakage current of the insulating material is increased, a drying belt is formed due to the thermal effect to generate partial arc, and finally the partial arc bridge initiates the overall flashover. From the above process, the surface wetting of the insulating material is a key factor causing the surface flashover, so the detection of the surface wettability of the insulating material is very important in the research and control of the mechanism of the surface flashover.
However, no specific method and evaluation index are formed in the aspect of online detection of the surface wettability of the insulating material. Three methods for detecting wettability are mainly proposed at present: the first is a leakage current method, which is to apply voltage to the high-voltage end of an insulating material and measure the leakage current flowing through the insulating material at the ground end, and judge the wettability by combining the leakage current and the fundamental component variation trend thereof, however, the leakage current value and the externally applied voltage have a remarkable relationship, the quantitative relationship between the unsaturated wettability and the leakage current value is difficult to form, and the tiny current measurement and the lack of wettability distribution information are also the main factors limiting the application of the method; the second method is a conductivity method, the surface conductivity is obviously increased along with the dissolution of soluble pollutants after the insulating material is wetted, and the surface conductivity can be calculated by the ratio of externally applied voltage to leakage current, however, the problem that the measurement of tiny current is difficult and the moisture degree distribution is not contained is also existed; in order to realize the detection of the wettability distribution, partial scholars propose a measurement method of local conductivity, and the wettability detection of a local area can be realized by applying direct-current voltage to two ends of an electrode with a specific structure and measuring leakage current between the electrodes, and the method can greatly improve the wettability measurement precision and obtain the wettability distribution information to a certain extent, but the contact measurement method has larger application limit; the third method is a mass measurement method, which realizes indirect measurement of wettability by measuring the mass of an insulating material under different wettability, but the method is also difficult to be applied on site. Therefore, it is necessary to provide a detection technology capable of realizing rapid measurement of the surface wettability of the insulating material and the distribution information thereof.
The above information disclosed in the background section is only for enhancement of understanding of the background of the invention and therefore may contain information that does not form the prior art that is already known to a person of ordinary skill in the art.
Disclosure of Invention
The invention aims to provide an in-situ measurement method for the surface wettability distribution of an insulating material based on spectrum inversion, which is used for establishing detection methods suitable for the surface wettability of different insulating materials by combining spectrum information with space image information, dynamically detecting a local drying zone caused by a thermal effect under an operating voltage, providing guidance for on-line detection of electric equipment and formulation of an operation and maintenance strategy of the electric equipment, and providing basis for design and dampproof design of insulating structures in different electric equipment. In order to achieve the above object, the present invention provides the following technical solutions:
the invention discloses an in-situ measurement method for the surface wettability distribution of an insulating material based on spectrum inversion, which comprises the following steps:
Step S1: acquiring surface wettability spectrum image data of an insulating material, wherein the insulating material of each sample is weighed under different wetting time, the mass of the insulating material at the current moment is recorded, each sample mass of the insulating material is converted into wettability Y, Y= |Y 1,Y2,...,Yn |, n is the number of wetting time intervals, and the surface wettability spectrum image data DN, DN= [ DN 1(x,y,λ),DN2(x,y,λ),...,DNn (x, Y, lambda) ] of each sample are synchronously acquired;
Step S2: the method comprises the steps of preprocessing surface wettability spectrum image data of an insulating material, wherein the different wettability spectrum image data DN n (x, y, lambda) is converted into different wettability reflectivity data S (lambda), S (lambda) = [ S 1(λ),S2(λ),...,Sn (lambda) ], wherein S n (lambda) is the average reflectivity value of the insulating material under different wettability, meanwhile, wavelength information (lambda) is sequenced from small to large, and the serial number of each lambda is recorded as a band serial number N. The reflectance data S (lambda) is subjected to outlier detection by a Monte Carlo sampling method, outliers are removed, and the average reflectance data of different wettabilities of group a is kept as X (lambda), and X (lambda) = [ X 1(λ),X2(λ),...,Xa (lambda) ]. Carrying out smoothing filtering, standard normalization, multi-component scattering correction, first derivative operation, second derivative operation and derivative operation data preprocessing on normal average reflectivity data X (lambda), recording the preprocessing result as X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (lambda), establishing a partial least square regression model based on wettability Y (a) = [ Y 1,Y2,...Ya ], normal average reflectivity data X (lambda) and preprocessing result X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (lambda), randomly selecting 75% of total samples as training sets in the modeling process, taking the remaining 25% as verification sets, introducing a decision coefficient R 2 and root mean square error RMSE as model evaluation indexes, and obtaining optimal wettability spectrum image data preprocessing results X' (lambda) of different insulating materials;
Step S3: extracting characteristic wave bands of regression analysis of the surface wettability spectrum image of the insulating material, wherein an optimal wettability spectrum image data preprocessing result X '(lambda) is respectively input into three characteristic wave band models of a continuous projection algorithm SPA, a competitive self-adaptive weighting algorithm CARS and a random frog-leaping algorithm RF, the number of the characteristic wave bands is K SPA,KCARS,KRF, K is less than N, and the hyperspectral data X' SPA(λK),X'CARS(λK),X'RF(λK after preprocessing-characteristic wave band extraction are correspondingly obtained;
Step S4: establishing a regression analysis model of the surface wettability spectrum image of the insulating material, respectively establishing a partial least square regression model based on the wettability Y and hyperspectral data X' SPA(λK),X'CARS(λK),X'RF(λK after preprocessing-characteristic wave band extraction, introducing a determination coefficient R 2 and a Root Mean Square Error (RMSE) as model evaluation indexes, and optimizing an optimal characteristic wave band extraction algorithm and a spectral image regression analysis model for detecting the surface wettability distribution of the insulating material through the maximum determination coefficient and the minimum root mean square error;
Step S5: the method comprises the steps of carrying out dynamic in-situ detection on the surface wettability of an insulating material under the operating voltage, carrying out in-situ detection on the wettability of the insulating material to be detected by adopting a hyperspectral imaging system to obtain reflectance data S ' n (X, Y, lambda) of the insulating material to be detected, respectively carrying out data preprocessing on the reflectance data S ' n (X, Y, lambda) of the insulating material to be detected to obtain normal reflectance data X ' (X, Y, lambda), extracting preprocessing-characteristic wave band hyperspectral data X ' (X, Y, lambda K) from the normal reflectance data X ' (X, Y, lambda) according to an optimal characteristic wave band extraction algorithm, wherein K represents a selected characteristic wave band sequence number, lambda K is the corresponding wavelength, substituting the preprocessing-characteristic wave band hyperspectral data X ' (X, Y, lambda K) into an insulating material surface wettability spectrum image regression analysis model to calculate the surface wettability Y ' (X, Y) of the insulating material to be detected, obtaining surface wettability detection results Y ' (X, Y, t) of the insulating material to be detected at different moments t, and carrying out visual processing on the surface wettability detection results Y ' (X, Y, t) to obtain the dynamic in-situ detection result of the operating voltage.
In the in-situ measurement method of the surface wettability distribution of the insulating material based on spectrum inversion, the step of converting the mass of the insulating material into the wettability Y comprises the following steps:
The method comprises the steps of weighing the mass of an insulating material when the insulating material is not wetted by a balance, recording the temperature and humidity in a fog chamber when the relative humidity in the fog chamber reaches 100%, placing an insulating material sample, taking out the sample after a certain time interval, weighing and recording the sample, wherein the mass is m 1, acquiring spectrum image data, weighing the sample again to be m 2 after the spectrum image data is acquired, and then placing the sample back to the fog chamber for carrying out the next group of tests, wherein the method for calculating the wettability of the insulating material comprises the following steps: y n＝((m1+m2)/2-m0)/m ', wherein Y n is the wettability of the insulating material at the nth moisture time interval, m 0 is the mass of the dry insulating material, m 1 is the mass of the insulating material before the acquisition of the spectrum image data, m 2 is the mass of the insulating material after the acquisition of the spectrum image data, and m' is the mass of the insulating material during saturated moisture.
In the in-situ measurement method of the surface wettability distribution of the insulating material based on spectrum inversion, the preprocessing steps of the surface wettability spectrum image data of the insulating material are as follows:
respectively acquiring response intensities DN white (x, y, lambda) of a standard white board under different wave bands and response intensities DN black (x, y, lambda) under different wave bands under a lens cover of a closed hyperspectral image acquisition system, and respectively averaging DN whie (x, y, lambda) and DN black (x, y, lambda) under all coordinate points to obtain: And Normalizing response intensity DN n (x, y, lambda) of the spectral image data of the insulating material to obtain S n (x, y, lambda),
Further averaging the reflectivity data of the points at positions (x, y) to obtain an average reflectivity value S n (lambda),/>And the average reflectivities at different wettabilities were recorded as S (λ), S (λ) = [ S 1(λ),S2(λ),...,Sn (λ) ], N represents the wetting time interval, λ represents the wavelength, band number N represents the wavelength λ is for the nth band in the detection range of the hyperspectral imaging system, i.e. n=1 when λ=400 nm, n=2 when λ=403.2 nm, …, n=176 when λ=900 nm.
For reflectivity data S (lambda), performing outlier detection by using a Monte Carlo sampling method, performing cross validation by using a partial least square method and a principal component regression method, determining the number of principal components, randomly dividing the whole data into a training set and a verification set by using Monte Carlo random sampling, predicting the training set to 75% of all data, obtaining the prediction error of each verification sample by using the verification set, performing loop execution 2500 times, finally obtaining the prediction error distribution of each sample, detecting outliers by using the statistical characteristics of the prediction error distribution, removing the outliers, and retaining normal average reflectivity data X (lambda), wherein a represents the number of different wettability data groups remained after the outlier detection.
Smoothing for each set of X a (λ) in the normal average reflectance data yields a preconditioning result X S-G,a (λ): the smoothing filter window is controlled to be 2c+1, and the original data points in the window are fitted by a k-1 th order polynomial, X S-G,a(λ)＝B·(BT·B)-1·BT·Xa (lambda), wherein,2C+1 denotes the total 2c+1 data points within the window, where c denotes the window coefficients, using 2c+1 to ensure that the number in the window is odd, k denotes the power of k of the polynomial fit,
Standard normalization was performed for each set of X a (λ) in the normal average reflectance data to yield the preconditioning result X SNV,a (λ): calculating to obtain average spectrum dataWherein i=1, 2, … N is the spectral band number; spectrum data standard deviation/>Standard normalization of X a (λ): X a (λ) represents the reflectance value at the ith band at the a-th wetting time, where a=1, 2 … a, representing the number of samples remaining after detection by outliers.
The multivariate scatter correction was performed for each set of X a (λ) in the normal average reflectance data to give a preconditioned result X MSC,a (λ): calculating to obtain average spectrum dataWherein j=1, 2, … a, a is the number of samples after the abnormal point is removed; obtaining/>, by linear regressionObtain the pretreatment result/>E j represents the fitting coefficients of the first order terms in the linear regression, and Q j represents the fitting coefficients of the constant terms in the linear regression;
Performing a first derivative operation for each set of X a (λ) in the normal average reflectance data yields a preconditioning result X FD,a (λ): wherein lambda N is a wavelength lambda corresponding to the nth band, and X a(λN) represents a reflectivity value of the nth band;
Performing a second derivative operation on each set of X a (λ) in the normal average reflectance data yields a preconditioning result X SD,a (λ):
Reciprocal operations are performed for each set of X a (λ) in the normal average reflectance data to yield a preconditioning result X RE,a (λ):
The X S-G,a(λ),XSNV,a(λ),XMSC,a(λ),XFD,a(λ),XSD,a(λ),XRE,a (lambda) set of data from group a is denoted as X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (lambda), respectively;
Establishing a partial least squares regression model based on the wettability Y (a) = [ Y 1,Y2,...Ya ], the average reflectivity data X (lambda) and the pretreatment result X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (lambda), randomly selecting 75% of total samples as training sets in the modeling process, taking the remaining 25% as verification sets, and introducing a decision coefficient R 2 and a Root Mean Square Error (RMSE) as model evaluation indexes: Where h 1,…,h7 is a partial least squares regression coefficient, l 1,…,l7 is a constant term, R 2 1,…,R2 7 is a determination coefficient, and RMSE 1,…,RMSE7 is a root mean square error. The calculation method of the decision coefficient R 2 is/> The calculation method of RMSE is/>Y mean (a) is the mean value of the wettability Y (a), and Y pre (a) is a predicted value of the wettability Y (a) based on the average reflectivity data X (λ) and the preprocessing result X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (λ) by establishing a partial least squares regression model. And obtaining the optimal wettability spectrum image data preprocessing result X' (lambda) of different insulating materials according to the maximum determination coefficient max { R 2 1,R2 2,..,R2 7 } and the minimum root mean square error min { RMSE 1,RMSE2,..,RME7 }.
In the in-situ measurement method of the surface wettability distribution of the insulating material based on spectrum inversion, the establishing step of the regression analysis model of the surface wettability spectrum image of the insulating material comprises the following steps:
Respectively establishing partial least squares regression models based on the wettability Y and hyperspectral data X' NSPA,X'NCARS,X'NRF after preprocessing-characteristic wave band extraction, and introducing a decision coefficient R 2 and a Root Mean Square Error (RMSE) as model evaluation indexes: Where h 1,…,h3 is a partial least squares regression coefficient, l 1,…,l3 is a constant term, R 2 1,…,R2 3 is a determination coefficient, and RMSE 1,…,RMSE3 is a root mean square error. Wherein, the calculation method of the determination coefficient R 2 is/> The calculation method of RMSE is/>Y mean (a) is the mean value of the wettability Y (a), and Y pre (a) is a predicted value of the wettability Y (a) based on the establishment of a partial least squares regression model. Selecting an optimal characteristic wave band sequence number K best for different insulating materials and an insulating material surface wettability spectrum image regression analysis model according to the principle of the maximum determination coefficient max { R 2 1,R2 2,R2 3 } and the minimum root mean square error min { RMSE 1,RMSE2,RMSE3 }: y=h ' X ' (λ Kbest) +l ', where h ' is a partial least squares regression coefficient and l ' is a constant term.
In the in-situ measurement method of the surface wettability distribution of the insulating material based on spectrum inversion, according to the optimal characteristic band wave band sequence number K best and a regression analysis model of the surface wettability spectrum image of the insulating material, the wettability of the surface of the insulating material at different moments Y' (x, Y, t) is calculated and pseudo-color processing is carried out, so that a visual result of dynamic in-situ detection of the surface wettability of the insulating material under the operating voltage is obtained.
In the technical scheme, the in-situ measurement method for the surface wettability distribution of the insulating material based on spectrum inversion has the following beneficial effects: according to the invention, target identification classification of different material components and structures is realized through spectrum analysis, and result visualization is realized by combining image information, and a detection method suitable for the surface wettability of different insulating materials is established by combining spectrum information and spatial image information, and dynamic detection is carried out on a local drying zone caused by thermal effect under the operating voltage.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments described in the present application, and other drawings may be obtained according to these drawings for a person having ordinary skill in the art.
FIG. 1 is a flow chart of a method for in situ measurement of the wettability distribution of an insulating material surface based on spectral inversion according to one embodiment of the present invention;
FIG. 2 is a schematic diagram of an artificial fog chamber of an insulation material surface wettability distribution in-situ measurement method based on spectral inversion according to one embodiment of the invention;
FIG. 3 is a hyperspectral imaging detection platform of an insulating material surface wettability distribution in-situ measurement method based on spectrum inversion according to the invention;
FIG. 4 is a flow chart of the wettability detection of an in-situ measurement method of the wettability distribution of the surface of an insulating material based on spectral inversion according to the present invention;
FIG. 5 is a screening result of Monte Carlo anomaly data for detecting the surface wettability of a silicone rubber material in a wettability detection application case of an insulation material surface wettability distribution in-situ measurement method based on spectral inversion according to the present invention;
FIG. 6 is a graph of different wettability spectra for the detection of the wettability of the surface of a silicone rubber material in the application case of the wettability detection in situ measurement method of the wettability distribution of the surface of an insulating material based on spectral inversion according to the present invention;
FIGS. 7 (a) to 7 (f) are the results of different pretreatment methods for detecting the surface wettability of the silicone rubber material in the application case of the wettability detection of the in-situ measurement method of the surface wettability distribution of the insulating material based on spectral inversion according to the present invention; fig. 7 (a) S-G filtering results, fig. 7 (b) SNV results, fig. 7 (c) MSC results, fig. 7 (d) FD results, fig. 7 (e) SD results, fig. 7 (f) RE results;
Fig. 8 (a) to 8 (c) are calculation results of different characteristic wave band extraction methods for detecting the surface wettability of the silicone rubber material in the application case of the wettability detection of the insulating material surface wettability distribution in-situ measurement method based on spectral inversion according to the present invention; FIG. 8 (a) SPA, FIG. 8 (b) CARS, FIG. 8 (c) RF;
Fig. 9 is a dynamic in-situ detection result for the formation process of a surface dry belt after the silicon rubber material is wetted in the application case of the in-situ measurement method of the surface wettability distribution of the insulating material based on the spectrum inversion according to the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, based on the embodiments of the invention, which are apparent to those of ordinary skill in the art without inventive faculty, are intended to be within the scope of the invention.
Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, based on the embodiments of the invention, which are apparent to those of ordinary skill in the art without inventive faculty, are intended to be within the scope of the invention.
It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further definition or explanation thereof is necessary in the following figures.
In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings are merely for convenience in describing the present invention and simplifying the description, and do not indicate or imply that the apparatus or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and thus should not be construed as limiting the present invention.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more such feature. In the description of the present invention, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.
In the present invention, unless explicitly specified and limited otherwise, the terms "mounted," "connected," "secured," and the like are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally formed; can be directly connected or indirectly connected through an intermediate medium, and can be communicated with the inside of two elements or the interaction relationship of the two elements. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art according to the specific circumstances.
In the present invention, unless expressly stated or limited otherwise, a first feature "above" or "below" a second feature may include both the first and second features being in direct contact, as well as the first and second features not being in direct contact but being in contact with each other through additional features therebetween. Moreover, a first feature being "above," "over" and "on" a second feature includes the first feature being directly above and obliquely above the second feature, or simply indicating that the first feature is higher in level than the second feature. The first feature being "under", "below" and "beneath" the second feature includes the first feature being directly under and obliquely below the second feature, or simply means that the first feature is less level than the second feature.
In order to make the technical scheme of the present invention better understood by those skilled in the art, the present invention will be further described in detail with reference to the accompanying drawings.
In one embodiment, as shown in fig. 1 to 9, the in-situ measurement method of the surface wettability distribution of the insulating material based on the spectral inversion comprises,
S1: spectral image data acquisition of surface wettability of insulating material
Insulating materials such as toughened glass, ceramic, high-temperature vulcanized silicone rubber and epoxy resin which are commonly used in the field of external insulation of an electric power system are selected, the insulating materials are subjected to artificial staining and then are subjected to wetting treatment in an artificial fog chamber, the insulating materials at the current moment are weighed and recorded under different wetting time, the insulating materials are converted into the wetting degree Y, Y= [ Y 1,Y2,...,Yn ], and n is the number of time intervals of a wetting test. And synchronously collecting hyperspectral original response data DN, DN= [ DN 1,DN2,...,DNn ] of each sample, wherein n is the number of the intervals of the wetting test duration. The wetting time may be any minute or hour as long as the same time interval is ensured. Thus in the following description, the nth dampening time interval or the like is used.
S2: pretreatment of spectral image data of surface wettability of insulating material
The obtained hyperspectral data DN is subjected to response intensity normalization and buried, and is converted into reflectivity data S, S= [ S 1,S2,...,Sn ], and for S n, the reflectivity data S is a three-dimensional matrix consisting of space image position information (x, y) and band sequence number (N), namely S n (x, y, N):
Because the data is abnormal due to the interference of external light or the specular reflection formed by a local water film area in the spectrum data acquisition process, in order to avoid the preferable influence of abnormal sample values on the characteristic wave band extraction and modeling, the abnormal values are firstly removed before the data analysis. And (3) performing outlier detection by applying a Monte Carlo sampling method on the reflectivity data S, and keeping the normal data as X.
Respectively carrying out data preprocessing such as smoothing filtering (S-G filtering), standard Normalization (SNV), multiple Scattering Correction (MSC), first derivative operation (FD), second derivative operation (SD), derivative operation (RE) and the like on the reflectivity data X, wherein the preprocessing results are recorded as follows: x S-G、XSNV、XMSC、XFD、XSD、XRE.
Based on the wettability Y, the normal data X and six preprocessing results X S-G、XSNV、XMSC、XFD、XSD、XRE, a partial least squares regression model is established, 75% of total samples are randomly selected as a training set in the modeling process, the remaining 25% are used as a verification set, and a decision coefficient R 2 and a Root Mean Square Error (RMSE) are introduced as model evaluation indexes, so that the optimal wettability spectrum image data preprocessing method under different insulating materials is optimized, and the optimal wettability spectrum image data preprocessing results X' of different insulating materials are obtained.
S3: characteristic wave band extraction for regression analysis of surface wettability spectrum image of insulating material
The optimal wettability spectrum image data preprocessing result X' is input into three characteristic wave band models of a continuous projection algorithm (SPA), a competitive self-adaptive weighting algorithm (CARS) and a random frog-leaping algorithm (RF) respectively, and the number of the characteristic wave bands is K SPA,KCARS,KRF and K is smaller than N.
And correspondingly obtaining hyperspectral data X' KSPA,X'KCARS,X'KRF after pretreatment-characteristic wave band extraction.
S4: establishment of regression analysis model of surface wettability spectrum image of insulating material
And respectively establishing partial least squares regression models based on the wettability Y and the hyperspectral data X' NSPA,X'NCARS,X'NRF after preprocessing-characteristic wave band extraction, and introducing a decision coefficient R 2 and Root Mean Square Error (RMSE) as model evaluation indexes, so that an optimal characteristic wave band extraction algorithm for different insulating materials and an insulating material surface wettability spectrum image regression analysis model are optimized.
S5: dynamic in-situ detection of surface wettability of insulating material under operating voltage
The hyperspectral imaging system is used for carrying out in-situ detection on hyperspectral image data of the insulating material to be detected to obtain reflectivity data S' (x, y, N) of the insulating material to be detected,
And carrying out in-situ detection on the wettability of the insulating material to be detected by adopting a hyperspectral imaging system, wherein the hyperspectral imaging system acquires image information of one slit width (namely one space dimension x) in each imaging period, and simultaneously disperses the image information into spectrum information (lambda) through a dispersion unit (prism, grating and prism-grating-prism), then controls the position of an entrance slit through a stepping motor, scans along the other space dimension (y) to obtain a three-dimensional space spectrum cube (x, y, lambda), and the wavelength lambda and the band sequence number N have a one-to-one correspondence to the hyperspectral imaging system.
Respectively preprocessing the reflectivity data S '(X, y, N) of the insulating material to be detected by using the surface wettability spectrum image data of the insulating material to obtain X' (X, y, N),
And extracting hyperspectral data under the characteristic wave bands from X '(X, y, N) according to an optimal characteristic wave band extraction algorithm of different insulating materials, wherein X' (X, y, K) is the hyperspectral data under the characteristic wave bands.
And substituting hyperspectral data X '(X, Y, N) under the characteristic wave band into an insulating material surface wettability spectrum image regression analysis model to calculate and obtain the surface wettability Y' (X, Y) of the insulating material to be detected.
And (3) carrying out in-situ detection on hyperspectral image data of the insulating material to be detected under the operating voltage by using a hyperspectral imaging system to obtain surface wettability detection Y' (x, Y, t) at different moments t. And (3) carrying out visualization treatment by using Y' (x, Y and t), thereby obtaining a dynamic in-situ detection result of the surface wettability of the insulating material under the operating voltage.
In a preferred embodiment of the method for in-situ measurement of the wettability distribution of the surface of an insulating material based on spectral inversion, the specific steps of converting the mass of the insulating material into the wettability Y are as follows:
The mass of the insulation material in the absence of moisture was weighed using a precision balance before the test began. And then opening the artificial fog chamber humidifying system, recording the temperature and humidity in the fog chamber when the relative humidity in the fog chamber reaches 100%, and placing an insulating material sample. And taking out the sample, weighing and recording the sample at a certain time interval, wherein the mass is m 1, acquiring hyperspectral image data, weighing the sample again to be m 2 after the hyperspectral image data are acquired, and then quickly placing the sample back to a fog room for the next group of tests. In the test process, the data acquisition speed is accelerated as much as possible, and the humidity of the sample surface is ensured to be less influenced by water evaporation by means of weighing twice and averaging. The method for calculating the wettability of the insulating material comprises the following steps: y i,n＝((m1+m2)/2-m0)/m'. Wherein Y i,n is the wettability of the insulating material, m 0 is the mass of the dry insulating material, m 1 is the mass of the insulating material before hyperspectral image acquisition, m 2 is the mass of the insulating material after hyperspectral image acquisition, and m' is the mass of the insulating material in saturated wetting.
In a preferred embodiment of the method for in-situ measurement of surface wettability distribution of an insulating material based on spectral inversion, the specific steps of preprocessing the surface wettability spectrum image data of the insulating material are as follows:
The response intensity is normalized, the response intensity DN white (lambda) of the standard white board under different wave bands and the response intensity DN black (lambda) under different wave bands under the condition that the lens cover of the hyperspectral image acquisition system is closed are respectively acquired, and then the following steps are obtained:
Abnormal point detection is carried out by applying a Monte Carlo sampling method to reflectivity data S, cross verification is carried out by adopting a partial least square method and a principal component regression method, and the number of principal components is determined. The whole data is then randomly divided into a training set and a validation set using monte carlo random sampling, the training set being between 75% of all data in size. And predicting by using a verification set, obtaining a prediction error of each verification sample, circularly executing 2500 times, finally obtaining a prediction error distribution of each sample, detecting abnormal points by using statistical characteristics of the prediction error distribution, removing the abnormal points, and keeping normal data as X, wherein X= [ X 1(x,y,N),X2(x,y,N),...,Xa (X, y, N) ], and a is the number of data points after the abnormal points are removed.
Smoothing for each set of X a (λ) in the normal average reflectance data yields a preconditioning result X S-G,a (λ): the smoothing filter window is controlled to be 2c+1, and the original data points in the window are fitted by a k-1 th order polynomial, X S-G,a(λ)＝B·(BT·B)-1·BT·Xa (lambda), wherein,2C+1 denotes the total 2c+1 data points within the window, where c denotes the window coefficients, using 2c+1 to ensure that the number in the window is odd, k denotes the power of k of the polynomial fit,
Standard normalization was performed for each set of X a (λ) in the normal average reflectance data to yield the preconditioning result X SNV,a (λ): calculating to obtain average spectrum dataWherein i=1, 2, … N is the spectral band number; spectrum data standard deviation/>Standard normalization of X a (λ): X a (λ) represents the reflectance value at the ith band at the a-th wetting time, where a=1, 2 … a, representing the number of samples remaining after detection by outliers.
The multivariate scatter correction was performed for each set of X a (λ) in the normal average reflectance data to give a preconditioned result X MSC,a (λ): calculating to obtain average spectrum dataWherein j=1, 2, … a, a is the number of samples after the abnormal point is removed; obtaining/>, by linear regressionObtain the pretreatment result/>E j denotes the fitting coefficient of the first order term in the linear regression, Q j denotes the fitting coefficient of the constant term in the linear regression,
Performing a first derivative operation for each set of X a (λ) in the normal average reflectance data yields a preconditioning result X FD,a (λ): Wherein lambda N is a wavelength lambda corresponding to the nth band, X a(λN) represents a reflectance value of the nth band,
Performing a second derivative operation on each set of X a (λ) in the normal average reflectance data yields a preconditioning result X SD,a (λ):
Reciprocal operations are performed for each set of X a (λ) in the normal average reflectance data to yield a preconditioning result X RE,a (λ):
The X S-G,a(λ),XSNV,a(λ),XMSC,a(λ),XFD,a(λ),XSD,a(λ),XRE,a (lambda) set of data for group a is denoted as X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (lambda) respectively,
Establishing a partial least squares regression model based on the wettability Y (a) = [ Y 1,Y2,...Ya ], the average reflectivity data X (lambda) and the pretreatment result X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (lambda), randomly selecting 75% of total samples as training sets in the modeling process, taking the remaining 25% as verification sets, and introducing a decision coefficient R 2 and a Root Mean Square Error (RMSE) as model evaluation indexes: Where h 1,…,h7 is a partial least squares regression coefficient, 1 1,…,l7 is a constant term, R 2 1,…,R2 7 is a determination coefficient, and RMSE 1,…,RMSE7 is a root mean square error. The calculation method of the decision coefficient R 2 is/> The calculation method of RMSE is/>Y mean (a) is the mean value of the wettability Y (a), and Y pre (a) is a predicted value of the wettability Y (a) based on the average reflectivity data X (λ) and the preprocessing result X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (λ) by establishing a partial least squares regression model. And obtaining the optimal wettability spectrum image data preprocessing result X' (lambda) of different insulating materials according to the maximum determination coefficient max { R 2 1,R2 2,..,R2 7 } and the minimum root mean square error min { RMSE 1,RMSE2,..,RMSE7 }.
In a preferred embodiment of the in-situ measurement method for the surface wettability distribution of the insulating material based on spectral inversion, the specific steps of the dynamic in-situ detection of the surface wettability of the insulating material under the operating voltage are as follows:
Performing in-situ detection on hyperspectral image data of the insulating material to be detected by using a hyperspectral imaging system to obtain reflectivity data S' (x, y, N) of the insulating material to be detected,
Respectively preprocessing the reflectivity data S '(X, y, N) of the insulating material to be detected by using the surface wettability spectrum image data of the insulating material to obtain X' (X, y, N),
And extracting hyperspectral data under the characteristic wave bands from X '(X, y, N) according to an optimal characteristic wave band extraction algorithm of different insulating materials, wherein X' (X, y, K) is the hyperspectral data under the characteristic wave bands.
And substituting hyperspectral data X '(X, Y, K) under the characteristic wave band into an insulating material surface wettability spectrum image regression analysis model to calculate and obtain the surface wettability Y' (X, Y) of the insulating material to be detected.
And (3) carrying out in-situ detection on hyperspectral image data of the insulating material to be detected under the operating voltage by using a hyperspectral imaging system to obtain surface wettability detection Y' (x, Y, t) at different moments t. And performing pseudo-color treatment by using Y' (x, Y, t) so as to obtain a visual result of dynamic in-situ detection of the surface wettability of the insulating material under the operating voltage.
On-site application verification
1. By using the method of the invention, the surface wettability of the silicone rubber material is detected, and the parameters are set as follows:
TABLE 1 MCS parameter settings
Fig. 5 monte carlo anomaly data screening results, through MCS screening, the anomaly 4 data points are removed, and the remaining 44 data samples are retained for subsequent analysis.
FIG. 6 different wettability spectra curves for the detection of the surface wettability of a silicone rubber material. The overall trend of the spectral reflectance curves of the samples with different wettabilities is consistent but there is still a certain difference, and at lower wettabilities there is an absorption peak at 400nm that increases with increasing wetting and shifts to the left, exhibits a pronounced reflectance characteristic near the 739nm band and drops slightly in the band after 739nm but still maintains a higher reflectance value due to the o—h bond resonance in moisture. And the spectral reflectance values show good negative correlation with wettability, as seen from the reflectance values.
Fig. 7 (a) to 7 (f) are the results of different pretreatment methods for the detection of the surface wettability of the silicone rubber material. FIG. 7 (a) S-G filtering results. Fig. 7 (b) SNV results. Fig. 7 (c) MSC results. FIG. 7 (d) FD results. Fig. 7 (e) SD results. FIG. 7 (f) RE results. And further establishing a partial least square model based on different pretreatment methods, wherein the results are shown in the following table.
Table 2 Effect of partial least squares model constructed by pretreatment algorithm
In the established PLSR model, the decision coefficients R2 in the training set and the validation set based on the raw spectral data are 0.9851 and 0.9870, respectively, indicating a high correlation between the reflectance spectral values and the wettability. After pretreatment, the modeling effect of the S-G and SNV models is similar to that of the original spectrum data model, the MSC and RE models are better than the original spectrum data model, and the FD and SD models are worse than the original spectrum models. The result shows that redundant information and interference information in the original data are removed to a certain extent after the MSC and RE are preprocessed, so that the reliability of the PLSR model is improved. By comparing the determining coefficients R 2 and the root mean square error RMSE of the model training set and the checking set, the RE model shows better modeling effect, so that the RE model is used as a pretreatment method for developing subsequent analysis for the silicon rubber material.
Fig. 8 (a) to 8 (c) are calculation results of different characteristic wave band extraction methods for detecting the surface wettability of the silicone rubber material in the application case of the wettability detection of the insulating material surface wettability distribution in-situ measurement method based on spectral inversion according to the present invention. FIG. 8 (a) SPA. Fig. 8 (b) CARS. Fig. 8 (c) RF. Characteristic wave bands selected by the SPA are 405.6nm, 532.5nm, 730.2nm, 765.3nm, 872.8nm, 912.9nm, 968.2nm, 971.9nm and 975.6nm respectively. Characteristic wave bands selected by the CARS of the self-adaptive weighting algorithm are 434.4nm, 440.9nm, 532.5nm, 582.6nm, 592.7nm, 616.4nm, 726.7nm, 790.1nm, 894.6nm, 920.2nm and 975.6nm in sequence. Characteristic wave bands selected by the random frog-leaping algorithm RF are 405.6nm, 415.2nm, 457.1nm, 532.5nm, 545.7nm, 681.4nm, 912.9nm, 957.1nm, 968.2nm and 983.1nm respectively.
The results of the partial least squares model established for the three characteristic bands are shown in table 3. Compared with an original RE model which does not extract the characteristic wave band, the PLSR model based on the SPA, CARS and RF methods has a more excellent fitting effect on the wettability of both the training set and the verification set, and the characteristic wave band extraction method can be used for greatly removing redundant spectral information and data dimension. Furthermore, the fitting effect is better than the other two methods compared to the RF method, with a higher R 2 and a lower RMSE.
TABLE 3 modeling effect based on different characteristic wavelength extraction methods
The established silicon rubber wettability detection model equation is shown in the formula:
The spectral reflectance value of each pixel point is substituted into the formula, so that a wettability calculation result of each pixel point in the image can be obtained, and according to the wettability value, the surface wettability distribution visualization of the insulating material can be realized by combining pseudo-color processing. Fig. 9 is a dynamic in-situ detection result for the formation process of a surface dry belt after the silicon rubber material is wetted in the application case of the in-situ measurement method of the surface wettability distribution of the insulating material based on the spectrum inversion according to the present invention.
The result shows that the method can effectively realize in-situ dynamic detection of the surface wettability of the insulating material.
Finally, it should be noted that: the described embodiments are intended to be illustrative of only some, but not all, of the embodiments of the present application and, based on the embodiments herein, all other embodiments that may be made by those skilled in the art without the benefit of the present disclosure are intended to be within the scope of the present application.
While certain exemplary embodiments of the present invention have been described above by way of illustration only, it will be apparent to those of ordinary skill in the art that modifications may be made to the described embodiments in various different ways without departing from the spirit and scope of the invention. Accordingly, the drawings and description are to be regarded as illustrative in nature and not as restrictive of the scope of the invention, which is defined by the appended claims.

Claims (5)

1. The in-situ measurement method for the surface wettability distribution of the insulating material based on spectrum inversion is characterized by comprising the following steps of:
Step S1: acquiring surface wettability spectrum image data of an insulating material, wherein the insulating material of each sample is weighed under different wetting time, the mass of the insulating material at the current moment is recorded, each sample mass of the insulating material is converted into wettability Y, Y= [ Y 1,Y2,...,Yn ], n is the number of intervals of the wetting time, and the surface wettability spectrum image data DN, DN= [ DN 1(x,y,λ),DN2(x,y,λ),...,DNn (x, Y, lambda) ] of each sample are synchronously acquired;
Step S2: preprocessing the surface wettability spectrum image data of the insulating material, wherein the different wettability spectrum image data DN n (X, Y, lambda) is converted into different wettability reflectivity data S (lambda), S (lambda) = [ S 1(λ),S2(λ),...,Sn (lambda) ], S n (lambda) is the average reflectivity value of the insulating material under different wettability, meanwhile, the wavelength information lambda is sequenced from small to large, the sequence number of each lambda is recorded as a band sequence number N, abnormal points of the data are removed through a Monte Carlo sampling method, the normal different wettability average reflectivity data of group a is reserved as X (lambda), X (lambda) = [ X 1(λ),X2(λ),...,Xa (lambda) ], smoothing filtering, standard normalization, multi-element scattering correction, first derivative operation, second derivative operation and preprocessing are respectively carried out on the normal average reflectivity data X (lambda), the preprocessing result is correspondingly recorded as X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (lambda), the optimal error model is established based on the wettability Y (a) = [ Y 1,Y2,...Ya ], the average reflectivity data X (lambda) is calculated by using the average coefficient X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE as the optimal error of the optimal training model, and the optimal error set is obtained by taking the optimal set of the optimal wettability data as the optimal training model;
Step S3: extracting characteristic wave bands of regression analysis of the surface wettability spectrum image of the insulating material, wherein an optimal wettability spectrum image data preprocessing result X '(lambda) is respectively input into three characteristic wave band models of a continuous projection algorithm SPA, a competitive self-adaptive weighting algorithm CARS and a random frog-leaping algorithm RF, the number of the characteristic wave bands is K SPA,KCARS,KRF, K is less than N, and the hyperspectral data X' SPA(λK),X'CARS(λK),X'RF(λK after preprocessing-characteristic wave band extraction are correspondingly obtained;
Step S4: establishing a regression analysis model of the insulating material surface wettability spectrum image, namely establishing a partial least square regression model of the basic dryness wettability Y and the hyperspectral data X' SPA(λK),X'CARS(λK),X'RF(λK after pretreatment-characteristic wave band extraction respectively, introducing a determination coefficient R 2 and a Root Mean Square Error (RMSE) as model evaluation indexes, and optimizing an optimal characteristic wave band extraction algorithm and a spectrum image regression analysis model for detecting the surface wettability distribution of the insulating material through the maximum determination coefficient and the minimum root mean square error;
Step S5: the method comprises the steps of carrying out dynamic in-situ detection on the surface wettability of an insulating material under the operating voltage, carrying out in-situ detection on the wettability of the insulating material to be detected by adopting a hyperspectral imaging system to obtain reflectance data S ' n (X, Y, lambda) of the insulating material to be detected, respectively carrying out data preprocessing on the reflectance data S ' n (X, Y, lambda) of the insulating material to be detected to obtain normal reflectance data X ' (X, Y, lambda), extracting preprocessing-characteristic wave band hyperspectral data X ' (X, Y, lambda K) from the normal reflectance data X ' (X, Y, lambda) according to an optimal characteristic wave band extraction algorithm, wherein K represents a selected characteristic wave band sequence number, lambda K is the corresponding wavelength, substituting the preprocessing-characteristic wave band hyperspectral data X ' (X, Y, lambda K) into an insulating material surface wettability spectrum image regression analysis model to calculate the surface wettability Y ' (X, Y) of the insulating material to be detected, obtaining surface wettability detection results Y ' (X, Y, t) of the insulating material to be detected at different moments t, and carrying out visual processing on the surface wettability detection results Y ' (X, Y, t) to obtain the dynamic in-situ detection result of the operating voltage.
2. The method for in-situ measurement of the wettability distribution of a surface of an insulating material based on spectral inversion according to claim 1, wherein the step of converting the mass of the insulating material into the wettability Y preferably comprises:
The method comprises the steps of weighing the mass of an insulating material when the insulating material is not wetted by a balance, recording the temperature and humidity in a fog chamber when the relative humidity in the fog chamber reaches 100%, placing an insulating material sample, taking out the sample after a certain time interval, weighing and recording the sample, wherein the mass is m 1, acquiring spectrum image data, weighing the sample again to be m 2 after the spectrum image data is acquired, and then placing the sample back to the fog chamber for carrying out the next group of tests, wherein the method for calculating the wettability of the insulating material comprises the following steps: y n＝((m1+m2)/2-m0)/m ', wherein Y n is the wettability of the insulating material at the nth moisture time interval, m 0 is the mass of the dry insulating material, m 1 is the mass of the insulating material before the acquisition of the spectrum image data, m 2 is the mass of the insulating material after the acquisition of the spectrum image data, and m' is the mass of the insulating material during saturated moisture.
3. The in-situ measurement method of the surface wettability distribution of the insulating material based on spectrum inversion according to claim 1, wherein the preprocessing step of the surface wettability spectrum image data of the insulating material is as follows: respectively acquiring response intensities DN white (x, y, lambda) of a standard white board under different wave bands and response intensities DN black (x, y, lambda) under different wave bands under a lens cover of a closed hyperspectral image acquisition system, and respectively averaging DN white (x, y, lambda) and DN black (x, y, lambda) under all coordinate points to obtain: And/> Normalizing response intensity DN n (x, y, lambda) of the spectral image data of the insulating material to obtain S n (x, y, lambda),Further averaging the reflectivity data of the points at positions (x, y) to obtain an average reflectivity value S n (lambda),/>And the average reflectance at different wettabilities was recorded as S (λ), S (λ) = [ S 1(λ),S2(λ),...,Sn (λ) ];
For reflectivity data S (lambda), performing outlier detection by using a Monte Carlo sampling method, performing cross validation by using a partial least square method and a principal component regression method, determining the number of principal components, randomly dividing the whole data into a training set and a verification set by using Monte Carlo random sampling, predicting the training set to 75% of all data, obtaining the prediction error of each verification sample by using the verification set, performing loop execution 2500 times, finally obtaining the prediction error distribution of each sample, detecting outliers by using the statistical characteristics of the prediction error distribution, removing the outliers, and retaining normal average reflectivity data X (lambda), wherein a represents the number of different wettability data sets remained after the outlier detection;
In the in-situ measurement method of the surface wettability distribution of the insulating material based on spectrum inversion, the preprocessing steps of the surface wettability spectrum image data of the insulating material are as follows:
Respectively acquiring response intensities DN white (x, y, lambda) of a standard white board under different wave bands and response intensities DN black (x, y, lambda) under different wave bands under a lens cover of a closed hyperspectral image acquisition system, and respectively averaging DN white (x, y, lambda) and DN black (x, y, lambda) under all coordinate points to obtain: And Normalizing response intensity DN n (x, y, lambda) of spectral image data of insulating material to obtain S n (x, y, lambda),/>Further averaging the reflectivity data of the points at positions (x, y) to obtain an average reflectivity value S n (lambda),And recording the average reflectivities at different wettabilities as S (λ), S (λ) = [ S 1(λ),S2(λ),...,Sn (λ) ], N represents the wetting time interval, λ represents the wavelength, band number N represents the wavelength λ is for the nth band in the hyperspectral imaging system detection range, i.e. n=1 when λ=400 nm, n=2 when λ=403.2 nm, …, n=176 when λ=900 nm;
For reflectivity data S (lambda), performing outlier detection by using a Monte Carlo sampling method, performing cross validation by using a partial least square method and a principal component regression method, determining the number of principal components, randomly dividing the whole data into a training set and a verification set by using Monte Carlo random sampling, predicting the training set to 75% of all data, obtaining the prediction error of each verification sample by using the verification set, performing loop execution 2500 times, finally obtaining the prediction error distribution of each sample, detecting outliers by using the statistical characteristics of the prediction error distribution, removing the outliers, and retaining normal average reflectivity data X (lambda), wherein a represents the number of different wettability data sets remained after the outlier detection;
smoothing for each set X a (λ) of normal average reflectance data yields a preconditioning result X S-G,a (λ): the smoothing filter window is controlled to be 2c+1, and the original data points in the window are fitted by a k-1 th order polynomial, X S-G,a(λ)＝B·(BT·B)-1·BT·Xa (lambda), wherein, 2C+1 denotes the total 2c+1 data points in the window, wherein c denotes the window coefficient, the number in the window is ensured to be odd by using 2c+1, and k denotes the k power of polynomial fitting;
Standard normalization was performed for each set of X a (λ) in the normal average reflectance data to yield the preconditioning result X SNV,a (λ): calculating to obtain average spectrum data Wherein i=1, 2, … N is the spectral band number; spectrum data standard deviation/>Standard normalization of X a (λ): X a (λ) represents the reflectance value at the ith band at the a-th wetting time, where a=1, 2 … a, represents the number of samples remaining after detection by outliers;
The multivariate scatter correction was performed for each set of X a (λ) in the normal average reflectance data to give a preconditioned result X MSC,a (λ): calculating to obtain average spectrum data Wherein j=1, 2, … a, a is the number of samples after the abnormal point is removed; obtaining/>, by linear regressionObtain the pretreatment result/>E j represents the fitting coefficients of the first order terms in the linear regression, and Q j represents the fitting coefficients of the constant terms in the linear regression;
Performing a first derivative operation on each set of X a (lambda) in the normal average reflectivity data to obtain a preprocessing result Wherein lambda N is a wavelength lambda corresponding to the nth band, and X a(λN) represents a reflectivity value of the nth band;
Performing a second derivative operation on each set of X a (λ) in the normal average reflectance data yields a preconditioning result X SD,a (λ):
Reciprocal operations are performed for each set of X a (λ) in the normal average reflectance data to yield a preconditioning result X RE,a (λ):
The X S-G,a(λ),XSNV,a(λ),XMSC,a(λ),XFD,a(λ),XSD,a(λ),XRE,a (lambda) set of data from group a is denoted as X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (lambda), respectively;
Establishing a partial least squares regression model based on the wettability Y (a) = [ Y 1,Y2,...Ya ], the average reflectivity data X (lambda) and the pretreatment result X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (lambda), randomly selecting 75% of total samples as training sets in the modeling process, taking the remaining 25% as verification sets, and introducing a decision coefficient R 2 and a Root Mean Square Error (RMSE) as model evaluation indexes: Wherein, h 1,…,h7 is a partial least squares regression coefficient, l 1,…,l7 is a constant term, R 2 1,…,R2 7 is a determination coefficient, and RMSE 1,…,RMSE7 is a root mean square error; the calculation method of the decision coefficient R 2 is/> The calculation method of RMSE is/>Y mean (a) is the average value of the wettability Y (a), Y pre (a) is the predicted value of the partial least squares regression model for the wettability Y (a) based on the average reflectivity data X (lambda) and the pretreatment result X S-G(λ)、XSNV(λ)、XMSC(λ)、XFD(λ)、XSD(λ)、XRE (lambda); and obtaining the optimal wettability spectrum image data preprocessing result X '(lambda) of different insulating materials according to the maximum determination coefficient max { R 2 1,R2 2,..,R2 7 } and the minimum root mean square error min { RM' SE 1,RMSE2,..,RMSE7 }.
4. The method for in-situ measurement of surface wettability distribution of an insulating material based on spectral inversion according to claim 1, wherein the step of establishing a regression analysis model of the surface wettability spectrum image of the insulating material comprises:
Respectively establishing partial least squares regression models based on the wettability Y and hyperspectral data X' NSPA,X'NCARS,X'NRF after preprocessing-characteristic wave band extraction, and introducing a decision coefficient R 2 and a Root Mean Square Error (RMSE) as model evaluation indexes: Wherein h 1,…,h3 is a partial least squares regression coefficient, l 1,…,l3 is a constant term, R 2 1,…,R2 3 is a determination coefficient, RMSE 1,…,RMSE3 is a root mean square error, and the calculation method of the determination coefficient R 2 is/> The calculation method of RMSE is/>Y mean (a) is the mean value of the wettability Y (a), and Y pre (a) is a predicted value of the wettability Y (a) based on the establishment of a partial least squares regression model; selecting an optimal characteristic wave band sequence number K best for different insulating materials and an insulating material surface wettability spectrum image regression analysis model according to the principle of the maximum determination coefficient max { R 2 1,R2 2,R2 3 } and the minimum root mean square error min { RMSE 1,RMSE2,RMSE3 }: y=h 'X' (λ 'Kbest) +l', where h 'is a partial least squares regression coefficient and l' is a constant term.
5. The in-situ measurement method for the surface wettability distribution of the insulating material based on the spectral inversion of claim 1 is characterized in that according to an optimal characteristic band sequence number K best and a regression analysis model of the surface wettability spectrum image of the insulating material, the wettability of the surface at different moments Y' (x, Y, t) is calculated and pseudo-color processing is carried out, so that a visual result of dynamic in-situ detection of the surface wettability of the insulating material under the operating voltage is obtained.
CN202210478805.1A 2022-04-28 2022-04-28 In-situ measurement method for surface wettability distribution of insulating material based on spectrum inversion Active CN115575331B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210478805.1A CN115575331B (en) 2022-04-28 2022-04-28 In-situ measurement method for surface wettability distribution of insulating material based on spectrum inversion

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210478805.1A CN115575331B (en) 2022-04-28 2022-04-28 In-situ measurement method for surface wettability distribution of insulating material based on spectrum inversion

Publications (2)

Publication Number Publication Date
CN115575331A CN115575331A (en) 2023-01-06
CN115575331B true CN115575331B (en) 2024-04-26

Family

ID=84579504

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210478805.1A Active CN115575331B (en) 2022-04-28 2022-04-28 In-situ measurement method for surface wettability distribution of insulating material based on spectrum inversion

Country Status (1)

Country Link
CN (1) CN115575331B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106124942A (en) * 2016-06-27 2016-11-16 华北电力大学（保定） A kind of based on infrared and ultraviolet image method zero resistance insulator detection method
BR102016005833A2 (en) * 2017-09-05 METHOD FOR PREDIZING SURFACE MOLHABILITY THROUGH MULTIVARIATE ANALYSIS ASSOCIATED WITH SPECTROSCOPIC METHODS
CN109164014A (en) * 2018-06-28 2019-01-08 浙江理工大学 A kind of multicolour cloth wetting zones recognition methods based on Hyperspectral imagery processing
CN109799442A (en) * 2019-03-29 2019-05-24 云南电网有限责任公司电力科学研究院 Insulator contamination prediction technique and system based on airborne hyperspectral
CN111239073A (en) * 2020-01-17 2020-06-05 西南交通大学 Composite insulator aging degree detection method based on hyperspectral technology

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
BR102016005833A2 (en) * 2017-09-05 METHOD FOR PREDIZING SURFACE MOLHABILITY THROUGH MULTIVARIATE ANALYSIS ASSOCIATED WITH SPECTROSCOPIC METHODS
CN106124942A (en) * 2016-06-27 2016-11-16 华北电力大学（保定） A kind of based on infrared and ultraviolet image method zero resistance insulator detection method
CN109164014A (en) * 2018-06-28 2019-01-08 浙江理工大学 A kind of multicolour cloth wetting zones recognition methods based on Hyperspectral imagery processing
CN109799442A (en) * 2019-03-29 2019-05-24 云南电网有限责任公司电力科学研究院 Insulator contamination prediction technique and system based on airborne hyperspectral
CN111239073A (en) * 2020-01-17 2020-06-05 西南交通大学 Composite insulator aging degree detection method based on hyperspectral technology

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
基于紫外成像技术的接触网绝缘子污秽状态检测;金立军;姜浩然;杨帆;段伟;;高电压技术;20151130(第11期);全文 *
基于高光谱技术的绝缘子污秽含水量检测;马欢;郭裕钧;张血琴;张晓青;李春茂;吴广宁;;高电压技术;20200430(第04期);全文 *

Also Published As

Publication number Publication date
CN115575331A (en) 2023-01-06

Similar Documents

Publication Publication Date Title
KR100398362B1 (en) Method and apparatus for measuring skin moisture by using near-infrared reflectance spectroscopy
Qin et al. Prediction of apple internal quality using spectral absorption and scattering properties
CN1276249C (en) Non-invasive in-vivo tissue classification using near-infrared measurements
US20160354015A1 (en) Non-invasive Measurement of Skin Thickness and Glucose Concentration with Raman Spectroscopy and Method of Calibration thereof
US6871169B1 (en) Combinative multivariate calibration that enhances prediction ability through removal of over-modeled regions
Li et al. Pears characteristics (soluble solids content and firmness prediction, varieties) testing methods based on visible-near infrared hyperspectral imaging
WO2010046968A1 (en) Diagnostic device and diagnostic method for concrete structure
CN110987846B (en) Nitrate concentration prediction method based on iPLS-PA algorithm
WO2021196416A1 (en) Optical fiber raman system and method for online detection of tungsten flotation concentrate grades
WO2007129648A1 (en) Method of estimating plant leaf water stress, device of estimating plant leaf water stress, and program of estimating plant leaf water stress
EP1883351A2 (en) Method for predicting a blood glucose level of a person
CN111999258B (en) Spectral baseline correction-oriented weighting modeling local optimization method
CN115575331B (en) In-situ measurement method for surface wettability distribution of insulating material based on spectrum inversion
CN116026780B (en) Method and system for online detection of coating moisture absorption rate based on series strategy wavelength selection
Abrahamsson et al. Scatter correction of transmission near-infrared spectra by photon migration data: Quantitative analysis of solids
WO2020223234A1 (en) Water reflection analysis of encapsulated photovoltaic modules
EP1999460A1 (en) Method and sensor for determining the hydrocarbon dew point in a gas
CN104697939B (en) The spectral measurement method and its device of the transformer insulating paper degree of polymerization
CN108267422B (en) Abnormal sample removing method based on near infrared spectrum analysis
DE102005063263A1 (en) Measurement result generating method for use in e.g. transmissive spectroscopic application, involves processing compensation function representing dependency of distribution of lights by temperature of diode, within signal processing
CN114088660A (en) Insulating paper water content evaluation method based on robust wavelength screening
Hao et al. Application of effective wavelength selection methods to determine total acidity of navel orange
Yeh et al. Determination of moisture profile in foods during microwave heating using VIS-NIR spectroscopy
CN116202991B (en) Online detection method and system for moisture absorption rate of tablet coating
JP7266632B2 (en) Inspection method and processing inspection device for battery electrode material

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
GR01 Patent grant
GR01 Patent grant