RU2498275C2 - Remote classification method of oil impurities on water surface - Google Patents

Abstract

FIELD: measurement equipment.

SUBSTANCE: for classification of oil impurities on water surface, test water surface is irradiated in UV-range on excitation wave length λ_e, intensity of fluorescent radiation I(λ₁), I(λ₂), I(λ₃), I(λ₄) from the test water surface is recorded in four narrow spectral ranges with centres on wave lengths λ₁, λ₂, λ₃, λ₄, which have been chosen from the condition of maximum distance between classes in two-dimensional space of classification criteria

and

Values K₁ and K₂ are found for the test water surface, and belonging of oil impurity to one of the classes is estimated as per penetration of the found values K₁ and K₂ for the test water surface to the area corresponding to that class in two-dimensional space of classification criteria.

EFFECT: invention allows performing classification as per four groups: water with different characteristics, protein or algae in water, crude oil, heavy oil products, and light cleaned oil products.

5 dwg, 2 tbl

Description

Technical field

The invention relates to measuring equipment and can be used for rapid identification of oil spills and oil products in marine, lake and river water areas.

State of the art

One of the most promising classes of sensors for remote operational monitoring of oil and oil product spills is laser fluorimeters, the operation of which is based on the registration of fluorescence radiation from the studied water surface (see, for example, [1]).

Laser fluorimeters can detect oil pollution and identify them, or at least classify them into three groups: light refined petroleum products (NP), crude oil and heavy petroleum products [1]. The main feature in this case is the fluorescence spectrum recorded by the receiver of the laser meter.

Known methods for the classification of oil pollution on the surface of the water [2 and 3], which are similar and prototype of the invention. The prototype method consists in the fact that the studied water surface is irradiated at an excitation wavelength in the ultraviolet range, the fluorescence radiation from the studied water surface is recorded, and oil pollution is classified by comparing the measured fluorescence emission spectra with the reference (previously recorded) emission spectra of the samples.

The disadvantage of the prototype method is the need to measure the fluorescence spectrum in a wide spectral range, which requires an expensive multi-channel receiving device.

Disclosure of invention

и

, находят величины K₁ и K₂ для исследуемой водной поверхности, и о принадлежности нефтяного загрязнения к одному из классов судят по попаданию найденных величин K₁ и K₂ для исследуемой водной поверхности в область, соответствующую этому классу в двумерном пространстве классифицирующих признаков.The objective of the invention is to measure the fluorescence spectrum in narrow spectral ranges without loss of quality classification of oil pollution. The effect is achieved by the fact that in the proposed method for the classification of oil pollution on the surface of the water, the studied water surface is irradiated in the ultraviolet range at the excitation wavelength λ _in , the fluorescence radiation intensity I (λ ₁ ), I (λ ₂ ), I (λ ₃ ) is recorded, I (λ ₄ ) from the studied water surface in four narrow spectral ranges with centers at wavelengths λ ₁ , λ ₂ , λ ₃ , λ ₄ selected from the condition of maximum distance between classes in the two-dimensional space of classifying features

and

, find the values of K ₁ and K ₂ for the studied water surface, and whether the oil pollution belongs to one of the classes is judged by the finding of the found values of K ₁ and K ₂ for the studied water surface in the region corresponding to this class in the two-dimensional space of classifying features.

The proposed method allows classification into four groups: water with different characteristics (surface not contaminated with oil products), protein or algae in water; raw oil; heavy petroleum products; light refined petroleum products.

The wavelengths of the centers of the spectral ranges λ ₁ , λ ₂ , λ ₃ , λ ₄ and specific class boundaries depend on the specific excitation wavelength λ _{in the} studied water area and are determined from the condition of the maximum probability of a correct classification.

List of figures

Figure 1 schematically shows a device that implements the proposed method.

Figure 2 and Figure 3 presents some types of water and oil products in the space of the selected classification features.

Figure 4 and Figure 5 shows the separation of classes using linear piecewise boundaries (h (K ₁ , K ₂ )) for excitation wavelengths of 226 and 337 nm, respectively.

The implementation of the invention

The device in figure 1 contains a source of ultraviolet radiation 1, irradiating the water surface at a wavelength of excitation λ _in ; photodetector 2 detecting fluorescence radiation from the water surface in four narrow spectral ranges (with centers at wavelengths λ ₁ , λ ₂ , λ ₃ , λ ₄ ); processing unit 3, which according to the measurement data determines the values of K ₁ and K ₂ for the investigated water surface and checks whether the obtained values of K ₁ and K ₂ fall into the region corresponding to one of the classes in the two-dimensional space of classification features.

The proposed method for the classification of oil pollution on the surface of the water is as follows.

An ultraviolet radiation source 1 (for example, a laser with an excitation wavelength of 266, or 308, or 337 nm — these wavelengths account for the largest number of measured fluorescence spectra of water, oil, and oil products known from the generally available scientific and technical literature) irradiate the studied water surface 4 on the excitation wavelength λ _in (for example, the radiation source 1 may be on an aircraft carrier).

Irradiation of the water surface is carried out vertically down. Photodetector 2 registers the fluorescence radiation intensity I (λ ₁ ), I (λ ₂ ), I (λ ₃ ), I (λ ₄ ) from four water surfaces under study in four narrow spectral ranges with centers at wavelengths λ ₁ , λ ₂ , λ ₃ , λ ₄ .

The signals from photodetector 2 enter the processing unit 3, into which class boundaries are entered (I - water with various characteristics (surface not contaminated with oil products), protein or algae in water; II - crude oil; III - heavy oil products, IV - light refined oil products) predetermined for the used excitation wave λ _in and the studied water area.

и

для исследуемой водной поверхности и проводят проверку попадания найденных значений K₁ и K₂ в область, соответствующую одному из классов в двумерном пространстве классифицирующих признаков.In processing unit 3, according to the measurement data, the quantities

and

for the investigated water surface and verify that the found values of K ₁ and K ₂ fall into the region corresponding to one of the classes in the two-dimensional space of classifying features.

When flying around the studied water area, the result of block 3 is an array of data on the classification of oil pollution (map of oil pollution).

At present, there are quite a lot of generally available experimental data on fluorescence spectra (in different spectral ranges of registration) of various oils, oil products and pure water for different water areas and for different excitation wavelengths of 266, 308 and 337 nm, etc. (see, for example, [1, 4-6]).

The spectral ranges of registration of fluorescent radiation (with centers at wavelengths λ ₁ , λ ₂ , λ ₃ , λ ₄ ) are selected according to the maximum distances between the classes in the two-dimensional space of the classification feature. The results of mathematical modeling show:

- for an excitation wavelength of 266 nm, the classification problem can be solved using spectral ranges with centers at wavelengths of 296, 306, 350, 367 nm. Thus, for the excitation wavelength of 266 nm, the classifying features are contrasts:

- for an excitation wavelength of 337 nm, the classification problem can be solved using spectral ranges with centers at wavelengths of 395, 402, 408, 411 nm. Thus, for the excitation wavelength of 337 nm, the classifying features are contrasts:

Figure 2 presents: 1 - a model sample of water, 2 - water of the Black Sea, 3 - protein dissolved in water, 4 - chlorella algae in water, 5 - chlorococcus algae in water, 6 - despicable and humic acids in water, 7 - Libyan oil in water No. 1, 5 - Shaim oil in water, 9 - oil in water, 10 - Libyan oil in water No. 2, 11 - fuel oil in water, 12 - diesel fuel in water, 13 - fuel for jet aircraft in water, 14 - kerosene in water.

Figure 3 presents: 1 - dissolved organic matter (DOM), 2 - water p. Don, 3 - water from Oakville Creek, 4 - water from Creek Tvelf-mail, 5 - Libyan oil, 6 - Esso oil (premium), 7 - Esso oil, 8 - hydrochloric oil, 9 - diesel, 10 - kerosene .

In mathematical modeling of the operation of the oil pollution classification method on the water surface, it was believed that due to noise and errors, the intensity measurements in the fluorescence spectra are distributed according to the normal law. It was believed that each class consists of many substances; each substance is determined by normally distributed classification features; distribution parameters are determined for each substance separately.

The resulting classes are separated using hyperplanes (see, for example, [7]). Class separation using linear piecewise boundaries (h (K ₁ , K ₂ )) is presented in FIG. 4 and FIG. 5 for excitation wavelengths of 226 and 337 nm, respectively. The superscripts I, II, III, IV indicate pairs of shared classes (I - water, protein or algae in water; II - crude oil; III - heavy petroleum products, IV - light refined petroleum products).

The probability was found of the correct classification of oil pollution into four groups: water, protein, or algae in water; raw oil; heavy petroleum products; light refined petroleum products.

Tables 1 and 2 show the results of mathematical modeling of the classification of oil pollution on the water surface by the proposed method. Mathematical modeling was carried out for excitation wavelengths of 337 nm and 266 nm and different values of the relative rms noise value of the recording equipment.

Table 1. Probabilities of correct classification for an excitation wavelength of 266 nm σ,% one 2 3 four 5 Substance Model water sample 1,000 0.958 0.888 0.841 0.821 Black Sea Water 1,000 1,000 1,000 0,999 0,992 Protein dissolved in water 1,000 0.988 0.912 0.821 0.746 Chlorella seaweed in water 0.985 0.856 0.759 0.700 0.657 Alga chlorococcus in water 1,000 1,000 1,000 0,994 0.976 Despicable and humic acids in water 1,000 1,000 1,000 1,000 1,000 Libyan oil in water 0,999 0.913 0.821 0.752 0.695 Shaim oil in water 1,000 0.963 0.889 0.815 0.758 Oil in water 1,000 0.956 0.861 0.767 0.698 Libyan oil in water 1,000 0,996 0.951 0.867 0.779 Fuel oil in water 0.987 0.872 0.771 0.721 0.674 Diesel in water 1,000 1,000 0,999 0.988 0.961 Jet fuel in water 0,998 0.925 0.833 0.772 0.729 Kerosene in the water 1,000 0.967 0.887 0.809 0.737

Table 2. Probabilities of correct classification for an excitation wavelength of 377 nm σ,% one 2 3 four 5 Substance Moat 1,000 0.987 0.935 0.868 0.819 Water of the Don River 1,000 1,000 1,000 0,999 0,994 Oakville Creek Water 1,000 1,000 0,994 0.974 0.944 Creek Water Twelf Email 1,000 1,000 1,000 0,998 0.988 Libyan oil 0,998 0.924 0.794 0.674 0.578 Esso Oil (Premium) 1,000 1,000 0,996 0.979 0.942 Esso Oil 1,000 1,000 0,999 0.983 0.955 Salt oil 1,000 0,996 0.956 0.875 0.778 Diesel fuel 1,000 0,997 0.970 0.924 0.890 Kerosene 0,997 0.919 0.811 0.726 0.643

From tables 1 and 2 it is seen that the developed classification method allows you to reliably classify oil pollution on the water surface.

For an excitation length of 266 nm with a measurement noise of σ = 1%, the probability of a correct classification is no worse than 0.985, and in most cases it is practically 1 (accurate to three decimal places).

For an excitation length of 337 nm with a measurement noise of σ = 1%, the probability of a correct classification is no worse than 0.997, and in most cases it is practically 1 (accurate to three decimal places).

Thus, the proposed remote method for the classification of oil pollution on the water surface, based on the registration of fluorescence radiation in four narrow spectral ranges, allows you to reliably classify oil pollution on the water surface.

Information sources

1. Mezheris R. Laser remote sensing. - M .: World. 1987, - 550 p.

2. Patent US 3899213. Airborne laser remote sensing system for the detection and identification of oil spills. Date of Patent Aug. 12, 1975. Int. Cl. G01T 1/169; G01N 21/38.

3. Patent RU 2233438. A method for remote detection and identification of objects of organic origin. The patent is valid on 08.26.2003. IPC G01N 21/64.

4. Taer Abd Deydan, Patsaeva S.V., Fadeev V.V., Yuzhakov V.I. Spectral features of fluorescence of oil products in films and in the volume of water. // Optics of the atmosphere and the ocean. 1994. V.7. Number 4. S.455-463.

5. Remote control of the upper ocean. / V.M. Orlov, I.V.Samokhvalov, M.L. Belov and others. Novosibirsk: Science. Sib. Department, 1991.149 s.

6. Glushkov S.M., Fadeev V.V., Filippova E.M., Chubarov V.V. Problems of laser fluorimetry of organic impurities in natural waters. // Optics of the atmosphere and the ocean. - 1994. - T.7, No. 4. - S. 464-473.

7. Ayvazyan S.A., Buchstaber V.M., Enyukov I.S., Meshalkin L.D. Applied statistics: Classification and reduction of dimension. - M.: Finance and statistics. 1989, - 607 p.

Claims (1)

A remote method for classifying oil contaminants on a water surface, including irradiating a test water surface with an excitation wavelength in the ultraviolet range, recording fluorescence radiation from a test water surface in a wide spectral range, and classifying oil contaminants based on a comparison of measured fluorescence radiation spectra with reference (pre-recorded) emission spectra of samples, characterized in that the surface of the water is irradiated in ultraviolet at the excitation wavelength λ _in , the fluorescence radiation intensity I (λ ₁ ), I (λ ₂ ), I (λ ₃ ), I (λ ₄ ) from the studied water surface is recorded in four narrow spectral ranges with centers at wavelengths λ ₁ , λ ₂ , λ ₃ , λ ₄ selected from the condition of the maximum distance between classes in the two-dimensional space of classifying features

and

, find the values of K ₁ and K ₂ for the studied water surface, and whether the oil pollution belongs to one of the classes is judged by the finding of the found values of K ₁ and K ₂ for the studied water surface in the region corresponding to this class in the two-dimensional space of classifying features.

Images