A device for detecting spectral components of oil-containing liquid

Abstract

The present invention provides a spectral component detection device for oil-containing liquids, which solves the technical problem of in-situ oil concentration detection in water that can identify oil types in order to conduct real-time monitoring of the oil concentration in the water around different offshore oil platforms due to the different oil types contained in the production wastewater of different offshore oil platforms. The device can be widely used in the field of spectral analysis of oil-containing liquids. Specifically, it includes a housing, a detection window is provided at one end of the housing, a detection platform is provided inside the housing, and a plurality of through holes are provided on the detection platform. The through holes are provided with LED light sources and photodetectors, the LED light sources are used to emit ultraviolet light, and the photodetectors are used to collect spectral data; an electronic control module is also provided in the housing, the electronic control module is electrically connected to the LED light source and the photodetector; a host computer is provided in the electronic control module, and the host computer stores an oil concentration spectral image recognition model for processing and analyzing the collected spectral data.

Description

Technical Field

[0001] The invention belongs to the technical field of spectrum analysis of oil-containing liquids, and in particular relates to a spectrum component detection device for oil-containing liquids. Background Art

[0002] Petroleum is an extremely important mineral resource in today's society, serving as an essential raw material for key industrial products such as fuels, lubricants, solvents, plastics, and synthetic rubber. As terrestrial oil resources continue to deplete, offshore oil production provides strong support for ensuring a stable supply of these resources. However, the oil production process generates a large amount of wastewater, which, after treatment, is discharged into the seawater. Substandard treatment of this wastewater and subsequent discharge into the marine environment poses a serious threat to species diversity. Because the oil content of wastewater from different offshore oil platforms varies, in-situ oil-in-water detection capable of oil identification is required to monitor oil concentrations in the waters surrounding these platforms in real time. Summary of the Invention

[0003] The purpose of the present invention is to solve the deficiencies of the above-mentioned technology and provide a device for detecting the spectral components of oil-containing liquids.

[0004] To this end, the present invention provides a spectral component detection device for oil-containing liquids, comprising a housing, a detection window being provided at one end of the housing, a detection platform being provided inside the housing, a plurality of through holes being provided on the detection platform, and an LED light source and a photodetector being provided in the through holes, wherein the LED light source is used to emit ultraviolet light, and the photodetector is used to collect spectral data;

[0005] An electronic control module is also provided in the shell, which is electrically connected to the LED light source and the photodetector; a host computer is provided in the electronic control module, which stores an oil concentration spectrum image recognition model for processing and analyzing the collected spectral data.

[0006] Furthermore, four through holes are provided, and a first LED light source, a first photodetector, a second LED light source and a second photodetector are provided in sequence.

[0007] Furthermore, the center wavelength of the first LED light source is 275nm, and the center wavelength of the second LED light source is 345nm; a filter is also provided in front of the lens of the first photodetector and the second photodetector, and the center wavelength of the filter of the first photodetector is 310nm, and the center wavelength of the filter of the second photodetector is 390nm.

[0008] The present invention provides a device for detecting the spectral components of oily liquids, which has the following beneficial effects:

[0009] This device detects oil concentration in water without sample pretreatment; the detected concentration is directly displayed on the instrument. It also establishes an oil-in-water identification model based on characteristic wavelengths, and constructs an interference correction model and calibration method, enabling measurement of oil concentrations in different types of water. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a structural diagram of the exterior of the present invention;

[0011] Figure 2 It is a left side view of the present invention;

[0012] Figure 3 yes Figure 2 Cross-section along AA;

[0013] Figure 4 yes Figure 2 Cross-section along the BB;

[0014] Figure 5 It is the control principle diagram of the present invention;

[0015] Figure 6 It is a three-dimensional fluorescence spectrum of a light oil sample;

[0016] Figure 7 It is the three-dimensional fluorescence spectrum of the medium oil sample;

[0017] Figure 8 It is a three-dimensional fluorescence spectrum of a heavy oil sample;

[0018] Figure 9 This is the three-dimensional fluorescence spectrum of the light oil sample after scattering is removed;

[0019] Figure 10 This is the three-dimensional fluorescence spectrum of the medium oil sample after scattering is removed;

[0020] Figure 11 It is the three-dimensional fluorescence spectrum of heavy oil sample after scattering is removed;

[0021] Figure 12 This is the flow chart of KNN algorithm selecting characteristic wavelength region;

[0022] Figure 13 It is the characteristic wavelength region selected by the KNN algorithm;

[0023] Figure 14 This is the SVM parameter optimization diagram based on the KNN algorithm wavelength selection method;

[0024] Figure 15 This is the image of the change of fluorescence intensity with temperature at excitation / emission wavelengths of 275nm / 310nm for light oil with a concentration of 10mg / L;

[0025] Figure 16 This is the image of the change of fluorescence intensity with temperature at excitation / emission wavelengths of 275nm / 310nm for light oil with a concentration of 50mg / L;

[0026] Figure 17 This is the image of the change of fluorescence intensity with temperature at excitation / emission wavelengths of 275nm / 310nm for light oil with a concentration of 100mg / L;

[0027] Figure 18 This is the image of the change of fluorescence intensity with temperature at excitation / emission wavelengths of 275nm / 310nm for light oil with a concentration of 150mg / L;

[0028] Figure 19 This is the image of the change of fluorescence intensity with temperature at excitation / emission wavelengths of 275nm / 390nm for a medium-quality oil with a concentration of 10mg / L;

[0029] Figure 20 This is the image of the change of fluorescence intensity with temperature at excitation / emission wavelengths of 275nm / 390nm for a medium-quality oil with a concentration of 50mg / L;

[0030] Figure 21 This is the image of the change of fluorescence intensity with temperature at excitation / emission wavelengths of 275nm / 390nm for a medium-quality oil with a concentration of 100mg / L;

[0031] Figure 22 This is the image of the change of fluorescence intensity with temperature at excitation / emission wavelengths of 275nm / 390nm for a medium-quality oil with a concentration of 150mg / L;

[0032] Figure 23 This is the image of the change of fluorescence intensity at 345nm / 390nm excitation / emission wavelength with temperature for heavy oil with a concentration of 10mg / L;

[0033] Figure 24 This is the image of the change of fluorescence intensity at 345nm / 390nm excitation / emission wavelength with temperature for heavy oil with a concentration of 50mg / L;

[0034] Figure 25 This is the image of the change of fluorescence intensity at 345nm / 390nm excitation / emission wavelength with temperature for heavy oil with a concentration of 100mg / L;

[0035] Figure 26 This is the image of the change of fluorescence intensity with temperature at 345nm / 390nm excitation / emission wavelength for heavy oil with a concentration of 150mg / L;

[0036] Markings in the figure: 1. Shell; 2. Detection window; 3. First LED light source; 4. Second LED light source; 5. Pressing ring; 6. Filter; 7. First photodetector; 8. Second photodetector; 9. Detection platform; 10. Electronic control module; 11. Watertight connector. DETAILED DESCRIPTION

[0037] The present invention is further described below with reference to the accompanying drawings and specific examples to facilitate understanding of the present invention. The methods used in the present invention are conventional methods unless otherwise specified; the raw materials and devices used are conventional commercially available products unless otherwise specified.

[0038] like Figure 1 As shown, the present invention provides a device for detecting the spectral composition of oil-containing liquids. The device comprises a housing 1, with a circular detection window 2 formed at one end and constructed with highly transparent glass to reduce light loss. A detection platform 9 is disposed within the housing 1, adjacent to the detection window 2. The detection platform 9 has four through-holes arranged in a 90-degree circular array, all of which are inclined toward the center of the detection window 2.

[0039] like Figure 2 As shown, a first LED light source 3, a first photodetector 7, a second LED light source 4, and a second photodetector 8 are sequentially disposed within the four through holes. The center wavelength of the first LED light source 3 is 275 nm, and the center wavelength of the second LED light source 4 is 345 nm. A filter 6 is also disposed in front of the lenses of the first photodetector 7 and the second photodetector 8. The center wavelength of the filter 6 of the first photodetector 7 is 310 nm, and the center wavelength of the filter 6 of the second photodetector 8 is 390 nm.

[0040] like Figure 3 As shown in FIG4 , the first LED light source 3, the first photodetector 7, the second LED light source 4 and the second photodetector 8 are all surrounded by a pressure ring 5 for fixing the position of the detection component to prevent shaking that causes the detection component lens to be unable to align with the detection window 2.

[0041] The system also includes an electronic control module 10, which is electrically connected to the detection assembly and a watertight connector 11. This connector can reduce the impact of a humid working environment on the detection results. The electronic control module 10 houses a host computer that stores an oil concentration spectral image recognition model, which is used to analyze the spectral images collected by the detection assembly and calculate the oil concentration of the detected liquid.

[0042] This device can realize the integrated monitoring of multiple oil types and is constructed by three sets of excitation / emission wavelengths: 275 / 310nm, 275 / 390nm, and 345 / 390nm. Figure 5As shown, the whole is controlled and started by the host computer. The first LED light source 3 and the second LED light source 4 emit ultraviolet light to illuminate the oil sample in the water in the sample pool. The fluorescent substance in the sample produces fluorescence. The emitted fluorescence passes through 310nm and 390nm filters for fluorescence band selection. Finally, it is acquired by the photodetector and transmitted to the host computer. The obtained data is processed by the host computer.

[0043] The steps for establishing the oil concentration spectrum image recognition model include:

[0044] 1. Collecting 3D fluorescence spectra of samples of different densities

[0045] Main instruments: F-320 fluorescence spectrophotometer, 10mm×10mm quartz fluorescence cuvette, ultrasonic cell disruptor, electronic balance, beaker, measuring cylinder, glass rod and other conventional laboratory chemical utensils.

[0046] Main reagents: light oil, medium oil, heavy oil, sodium lauryl sulfate, ultrapure water.

[0047] Fluorescence spectrophotometer parameters: excitation wavelength: 200-450 nm; excitation / emission slit: 5 nm; scanning speed: 1200 nm·min -1 ; Scanning band: 250-500nm; Voltage: 700V; Scan the spectrum for automatic instrument calibration.

[0048] Sample preparation: 100 mg of each of the three crude oils was added to 1 L of ultrapure water to create a 100 mg / L oily water sample. Oily water samples of varying concentrations were obtained by adding 10 mL of the oily water sample to varying volumes of ultrapure water, as shown in Table 1. The light, medium, and heavy oil samples required for the experiment were all crude oil samples provided by offshore oil platforms.

[0049] Table 1 Concentration gradient setting of oily water samples

[0050] Oily water sample (mL) Ultrapure water (mL) Sample concentration (mg / L) 10 0 100.00 10 1 90.91 10 2 83.33 10 3 76.92 10 4 71.43 10 5 66.67 10 6 62.50 10 8 55.56 10 10 50.00 10 12 45.45 10 15 40.00 10 18 35.71 10 23 30.30 10 30 25.00 10 40 20.00 10 50 16.67 10 60 14.29 10 70 12.50 10 80 11.11 10 90 10.00

[0051] like Figure 6 、 7 Figures 8 and 8 show three-dimensional fluorescence spectra collected using an F-320 fluorescence spectrophotometer. Light oil has a main excitation wavelength of 250-310 nm and an emission wavelength of 310-370 nm; medium oil has an excitation wavelength of 250-380 nm and an emission wavelength of 330-420 nm; and heavy oil has an excitation wavelength of 260-400 nm and an emission wavelength of 340-430 nm. The peaks for light, medium, and heavy oils gradually shift toward longer wavelengths.

[0052] The present invention is applied in the waters near offshore oil platforms, where the wastewater discharged into the seawater is oily wastewater that has been treated by emulsification and other steps. Oil is a water-insoluble substance, so to prepare an oil-containing solution of a certain concentration in the laboratory, an emulsifier must be added to evenly disperse the oil in the water. Sodium dodecylbenzenesulfonate and sodium dodecyl sulfate are two of the more commonly used emulsifiers. Since the three-dimensional fluorescence spectra of different oils need to be detected, the emulsifier used to evenly disperse the oil in the water cannot have a fluorescent effect and cannot affect the fluorescence spectrum of the oil.

[0053] 2. Removing scattering from three-dimensional fluorescence spectra

[0054] When analyzing oil samples in water based on the light scattering effect, , Rayleigh scattering will appear in the spectrum; when , secondary Rayleigh scattering will appear in the spectrum. The frequency of light waves changes after being scattered, a phenomenon known as Raman scattering. Therefore, Raman scattering spectra will also appear in the spectrum. The fluorescence intensity can reach 106-108 times the intensity of Raman scattered light, so the Raman scattered light can be ignored. The three-dimensional fluorescence spectrum acquisition area in the present invention is not affected by secondary Rayleigh scattering. Therefore, the scattering treatment in the present invention focuses on primary Rayleigh scattering.

[0055] Since the fluorescence at the characteristic wavelength selected by the present invention needs to avoid the Rayleigh scattering interference area, the present invention directly deducts the Rayleigh scattering and uses the drEEM toolbox in MATLAB to remove the Rayleigh scattering. Figure 6 、 7 , 8. The Rayleigh scattering in the three-dimensional fluorescence spectra of light oil, medium oil and heavy oil is subtracted, such as Figure 9 、 10 , 11 show the Rayleigh scattering removal results of the three-dimensional fluorescence spectra of light oil, medium oil and heavy oil.

[0056] 3. Using KNN algorithm to select characteristic wavelengths of three-dimensional fluorescence spectra

[0057] The sample set is randomly divided into 80% training set and 20% validation set. The full spectrum data of the training set is used to establish the KNN model, and the characteristic wavelength region with the highest accuracy in the validation set is selected. Figure 12 The figure shows the flow chart for selecting characteristic wavelength regions. The wavelength groups AXY / BXY / CXY are the wavelengths selected for the fluorescence spectrum regions of light oil, medium oil, and heavy oil, respectively. Different wavelength combinations are trained using the KNN algorithm. The test set is placed in the model for testing to obtain the oil classification accuracy under the wavelength combination. All wavelength combinations are traversed until the classification accuracy under all wavelengths is obtained. The region with the higher accuracy is the characteristic wavelength selection region.

[0058] like Figure 13 The figure shows the characteristic wavelength regions selected using the KNN model. The regions with higher accuracy in the figure are: light oil excitation wavelength 260-280nm, emission wavelength 300-320nm; medium oil excitation wavelength 270-290nm, emission wavelength 370-390nm; heavy oil excitation wavelength 330-350nm, emission wavelength 390-440nm.

[0059] A linear regression model is established based on the characteristic wavelength region to further screen the characteristic wavelength with higher accuracy. In the linear regression model, R 2 The correlation coefficient represents the data. A higher correlation coefficient indicates better concentration linearity at that wavelength. Tables 3-2, 3-3, and 3-4 show the top ten wavelengths with the best linear regression model indicators for the three oils in the training set.

[0060] Table 2 Top ten wavelengths with the best linear regression model indicators for light oil

[0061] Serial number Excitation wavelength (nm) Emission wavelength (±10nm) <![CDATA[R 2 ]]> RMSEP 1 275 310 0.9801 0.0335 2 275 325 0.9784 0.0355 3 275 320 0.9764 0.0351 4 280 315 0.9743 0.0386 5 280 325 0.9741 0.0301 6 280 330 0.9733 0.0398 7 260 325 0.9715 0.0339 8 260 310 0.9713 0.0375 9 260 315 0.9639 0.0395 10 260 320 0.9626 0.0405

[0062] Table 3 Top ten wavelengths with the best linear regression model indicators for medium oil

[0063] Serial number Excitation wavelength (nm) Emission wavelength (±10nm) <![CDATA[R 2 ]]> RMSEP 1 280 380 0.9997 0.0105 2 280 385 0.9997 0.0102 3 280 400 0.9997 0.0093 4 280 385 0.9997 0.0108 5 285 395 0.9996 0.0133 6 290 380 0.9996 0.0129 7 285 390 0.9996 0.0117 8 285 385 0.9994 0.0109 9 270 385 0.9993 0.0135 10 285 400 0.9992 0.0147

[0064] Table 4 The top ten wavelengths with the best indicators in the heavy oil linear regression model

[0065] Serial number Excitation wavelength (nm) Emission wavelength (±10nm) <![CDATA[R 2 ]]> RMSEP 1 345 390 0.9973 0.0195 2 345 405 0.9973 0.0186 3 345 420 0.9973 0.0199 4 350 390 0.9972 0.0201 5 350 410 0.9972 0.0215 6 350 425 0.9971 0.0226 7 350 430 0.9971 0.0234 8 325 405 0.9971 0.0234 9 325 420 0.9968 0.0268 10 325 410 0.9968 0.0275

[0066] Tables 2, 3, and 4 show that within the characteristic wavelength region, the linear fits for each characteristic wavelength vary very little. The optimal excitation wavelength for light oil is 275 nm, and the optimal emission wavelength is 310 nm; the optimal excitation wavelength for medium oil is 280 nm, and the optimal emission wavelength is 380 nm; and the optimal excitation wavelength range for heavy oil is 345 nm, and the optimal emission wavelength range is 390 nm. Because the optimal emission wavelengths for medium and heavy oils are similar, and the difference in concentration detection accuracy between the optimal emission wavelengths of 380 nm and 390 nm for medium oil is minimal, three excitation / emission wavelength sets of 275 / 310 nm, 275 / 390 nm, and 345 / 390 nm were selected to classify the three oils, while ensuring concentration detection accuracy and maintaining the advantages of sensor miniaturization and low power consumption.

[0067] Based on the three excitation / emission wavelengths of 275 / 310 nm, 275 / 390 nm, and 345 / 390 nm selected by the KNN algorithm, the SVM algorithm was used to establish an oil in water classification model, and the classification accuracy of light oil, medium oil, and heavy oil samples was verified using this model. Figure 14The figure shows the SVM model parameter selection results. The x-axis is the logarithm of the C parameter with base 2, the y-axis is the logarithm of the gamma parameter with base 2, and the z-axis is the cross-validation accuracy of the classification model under the characteristic wavelength. When the C parameter is 0.25 and the gamma parameter is 32, the model has the highest cross-validation accuracy, reaching 100.00%.

[0068] 4. Temperature compensation of fluorescence intensity values ​​of three-dimensional fluorescence spectra

[0069] For light crude oil, take 10 mg, 50 mg, 100 mg, and 150 mg respectively and put them into a 1L beaker. Add 10 g of sodium lauryl sulfate and 1L of ultrapure water to each beaker to prepare oil samples with concentrations of 10 mg / L, 50 mg / L, 100 mg / L, and 150 mg / L.

[0070] For medium oil and heavy oil crude oil, take 10 mg, 50 mg, 100 mg, and 150 mg respectively and put them into a 1L beaker, add 10 g of sodium lauryl sulfate and 1L of ultrapure water to each, and shake with an ultrasonic cell disruptor for 1-2 hours to evenly disperse the medium oil and heavy oil in the water to prepare oil samples with concentrations of 10 mg / L, 50 mg / L, 100 mg / L, and 150 mg / L.

[0071] Temperature experiments were conducted on prepared oil samples at four concentrations: 10mg / L, 50mg / L, 100mg / L, and 150mg / L. The surface water temperature in my country's seas varied between 0°C and 30°C. However, due to the extremely low solubility of sodium lauryl sulfate below 5°C, white flocculent material appeared in the prepared samples, affecting fluorescence detection. Therefore, the temperature range for this experiment was set to 5°C to 35°C.

[0072] Place the prepared oil-containing sample in a constant temperature bath, and measure the fluorescence intensity of the oil-containing sample at around 5℃, 10℃, 15℃, 20℃, 25℃, 30℃, and 35℃, respectively. Measure three sets of data at each temperature and take the average value.

[0073] In the range of 5℃-35℃, the fluorescence intensity of light oil at concentrations of 10mg / L, 50mg / L, 100mg / L and 150mg / L was measured with a fluorescence spectrophotometer to determine the relationship between the temperature and the fluorescence intensity. Figure 15 、 16 Figures 17 and 18 show the trend of fluorescence intensity variation with temperature at 275nm / 310nm excitation / emission wavelength for four concentrations of oil-containing samples. As can be seen from the figure, the fitting function of the fluorescence intensity variation with temperature at a concentration of 10 mg / L is y=-1.0053x+376.9359, and the correlation coefficient R 2is 0.9350; the fitting function of the fluorescence intensity change trend with temperature at 50 mg / L concentration is y=-8.1724x+1509.5, and the correlation coefficient R 2 is 0.9498; the fitting function of the fluorescence intensity change trend with temperature at a concentration of 100 mg / L is y=-13.055x+2551.2, and the correlation coefficient R 2 is 0.9426; the fitting function of the fluorescence intensity change trend with temperature at a concentration of 150 mg / L is y=-26.458x+3586.9, and the correlation coefficient R 2 Therefore, regardless of low or high concentration, the fluorescence intensity of the light oil sample decreases with increasing temperature, and the slope of the fitting curve increases with increasing concentration.

[0074] In the range of 5℃-35℃, the fluorescence intensity of the medium oil at the concentrations of 10mg / L, 50mg / L, 100mg / L and 150mg / L was measured by fluorescence spectrophotometer. Figure 19 、 20 Figures 21 and 22 show the variation trend of fluorescence intensity with temperature at 275nm / 390nm excitation / emission wavelength for four concentrations of oil-containing samples. As can be seen from the figure, the fitting function of the variation trend of fluorescence intensity with temperature at a concentration of 10 mg / L is y=-5.6528x+851.9739, and the correlation coefficient R 2 is 0.9104; the fitting function of the fluorescence intensity change trend with temperature at 50 mg / L concentration is y=-14.9692x+2644.5, and the correlation coefficient R 2 is 0.9150; the fitting function of the fluorescence intensity change trend with temperature at a concentration of 100 mg / L is y=-23.373x+4170.2, and the correlation coefficient R 2 is 0.9158; the fitting function of the fluorescence intensity change trend with temperature at a concentration of 150 mg / L is y=-28.964x+5353.7, and the correlation coefficient R 2 Therefore, regardless of low or high concentration, the fluorescence intensity of the medium oil sample also decreases with increasing temperature, and the slope of the fitting curve also increases with increasing concentration.

[0075] In the range of 5℃-35℃, the fluorescence intensity of heavy oil at concentrations of 10mg / L, 50mg / L, 100mg / L and 150mg / L was measured using a fluorescence spectrophotometer. Figure 23 、 24Figures 25 and 26 show the trend of fluorescence intensity variation with temperature at 345nm / 390nm excitation / emission wavelength for four concentrations of oil-containing samples. As can be seen from the figure, the fitting function of the fluorescence intensity variation with temperature at a concentration of 10 mg / L is y=-2.4287x+709.2739, and the correlation coefficient R 2 is 0.9677; the fitting function of the fluorescence intensity change trend with temperature at 50 mg / L concentration is y=-15.47x+3744.1, and the correlation coefficient R 2 is 0.9293; the fitting function of the fluorescence intensity change trend with temperature at a concentration of 100 mg / L is y=-22.116x+5054.1, and the correlation coefficient R 2 is 0.9652; the fitting function of the fluorescence intensity change trend with temperature at a concentration of 150 mg / L is y=-32.058x+6501.6, and the correlation coefficient R 2 Therefore, regardless of low or high concentration, the fluorescence intensity of heavy oil samples decreases with increasing temperature, and the slope of the fitting curve increases with increasing concentration.

[0076] According to the law of the influence of temperature on the fluorescence intensity value, it can be found that the fluorescence intensity value decreases with increasing temperature and presents a linear relationship. The formula is:

[0077] ;

[0078] Where, represents the fluorescence intensity value; T represents the temperature, °C; represents the slope of the fitting curve between fluorescence intensity and temperature; Represents the intercept of the fitting curve. At the same time, the slope of the fitting curve increases with the increase of oil concentration, and also shows a linear relationship. The formula is:

[0079] ;

[0080] Where, represents the slope of the fitting curve with the oil concentration in water; C represents the oil concentration in water, mg / L; Represents a constant. Combining the above formula, we can get the change of fluorescence intensity with temperature as follows:

[0081] ;

[0082] When the temperature is 0℃, the fluorescence intensity is , the fluorescence intensity value that needs to be compensated under the influence of temperature is:

[0083] ;

[0084] Where, Indicates the fluorescence intensity value that needs to be compensated under the influence of temperature; represents the fluorescence intensity value at 0°C; represents the fluorescence intensity value at i℃; T represents temperature, ℃.

[0085] In the description of the present invention, it should be understood that the terms "left", "right", "up", "down", "top", "bottom", "front", "back", "inside", "outside", "back", "middle", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0086] However, the above description is merely a specific embodiment of the present invention and should not be used to limit the scope of implementation of the present invention. Therefore, the replacement of equivalent components, or equivalent changes and modifications made according to the scope of protection of the present invention should still fall within the scope covered by the claims of the present invention.

Claims

1. A device for detecting the spectral components of an oily liquid, comprising a housing, one end of which is provided with a detection window, characterized in that: A detection platform is provided inside the shell, and a plurality of through holes are opened on the detection platform. An LED light source and a photodetector are provided in the through holes. The LED light source is used to emit ultraviolet light, and the photodetector is used to collect spectral data. An electric control module is also provided in the housing, which is electrically connected to the LED light source and the photodetector; a host computer is provided in the electric control module, which stores an oil concentration spectrum image recognition model for processing and analyzing the collected spectrum data; The steps of establishing the oil concentration spectrum image recognition model include: S10: Collect three-dimensional fluorescence spectra of samples with different densities; S20: Select characteristic wavelengths of three-dimensional fluorescence spectra using the KNN algorithm; S30: Use the SVM algorithm combined with characteristic wavelengths to establish an oil-in-water classification model; S40: performing temperature compensation on the oil-in-water classification model to obtain an oil concentration spectrum image recognition model; The steps of performing temperature compensation on the oil-in-water classification model to obtain the oil concentration spectrum image recognition model include: Light oil, medium oil, and heavy oil were added with different volumes of pure water and emulsifier to obtain oily water samples of different concentrations. The water samples were placed in different ambient temperatures, and fluorescence intensity data of oil-containing water samples with different concentrations were collected; Based on the fluorescence intensity data collected from each oil-containing water sample, fluorescence intensity-temperature relationship images were established respectively; According to the changing trend of the fluorescence intensity-temperature relationship image, the fitting functions of different oil-containing water samples were established respectively; The temperature compensation model was established by integrating the fitting functions of different oily water samples; Temperature compensation of the oil-in-water classification model using a temperature supplement model; The steps of establishing a temperature compensation model by integrating the fitting functions of different oil-containing water samples include: According to the law of the influence of temperature on fluorescence intensity, the fluorescence intensity decreases with increasing temperature and presents a linear relationship. The formula is: ； Where, represents the fluorescence intensity value; T represents the temperature, °C; represents the slope of the fitting curve between fluorescence intensity and temperature; represents the intercept of the fitting curve; at the same time, the slope of the fitting curve increases with the increase of oil concentration, and also shows a linear relationship. The formula is: ； Where, represents the slope of the fitting curve with the oil concentration in water; C represents the oil concentration in water, mg / L; represents a constant; combined with the above formula, the change of fluorescence intensity with temperature can be obtained as follows: ； When the temperature is 0℃, the fluorescence intensity is , the fluorescence intensity value that needs to be compensated under the influence of temperature is: ； Where, Indicates the fluorescence intensity value that needs to be compensated under the influence of temperature; represents the fluorescence intensity value at 0°C; represents the fluorescence intensity value at i℃; T represents temperature, ℃.

2. The spectral component detection device for oil-containing liquid according to claim 1, characterized in that: Four through holes are provided, and a first LED light source, a first photodetector, a second LED light source and a second photodetector are provided in sequence.

3. The spectral component detection device for oil-containing liquid according to claim 2, characterized in that: The center wavelength of the first LED light source is 275nm, and the center wavelength of the second LED light source is 345nm; filters are also arranged in front of the lenses of the first photodetector and the second photodetector, and the center wavelength of the filter of the first photodetector is 310nm, and the center wavelength of the filter of the second photodetector is 390nm.

4. The spectral component detection device for oil-containing liquid according to claim 1, characterized in that: The steps for acquiring three-dimensional fluorescence spectra of samples of different densities include: Light oil, medium oil, and heavy oil were added with different volumes of pure water and emulsifier to obtain oily water samples of different concentrations. Three-dimensional fluorescence spectra were collected for oily water samples of different concentrations.

5. The spectral component detection device for oil-containing liquid according to claim 1, characterized in that: After collecting three-dimensional fluorescence spectra of samples with different densities, the scattering of the three-dimensional fluorescence spectra is removed to obtain pure three-dimensional fluorescence spectra.

6. The spectral component detection device for oil-containing liquid according to claim 1, characterized in that: After using the KNN algorithm to select the characteristic wavelengths of the three-dimensional fluorescence spectrum, a linear regression model was established based on the characteristic wavelengths to screen the characteristic wavelengths with high accuracy.

Images