Sea surface oil film thickness measurement method based on optimized laser-induced fluorescence spectrum wave band

Abstract

The invention provides a sea surface oil film thickness measuring method based on optimized laser-induced fluorescence spectrum wave bands, and belongs to the technical field of spectral analysis. The method comprises the following steps: emitting an excitation light beam to a to-be-measured sea surface oil film by using a laser, and exciting the oil film to generate a fluorescence signal; receiving the fluorescence signal based on a spectrum acquisition device, and obtaining fluorescence spectrum data of the oil film; selecting a characteristic wave band from the fluorescence spectrum data through a sparse partial least square model, and predicting the thickness of the sea surface oil film based on the characteristic wave band in combination with Lasso regression; meanwhile, the selection of characteristic wave bands is optimized by using a hedera helix algorithm. According to the method, the characteristic wave band of oil film detection is selected through the sparse partial least square model, the oil film measurement characteristic wave band is efficiently and accurately obtained, full-wave band collection is avoided, and the detection efficiency is improved; and in combination with the hedera helix optimized sparse partial least square model, the accuracy of characteristic wave band selection and the generalization of the model are further improved, and the precision and efficiency of oil film thickness measurement are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of spectrum analysis, in particular to a method for measuring the thickness of sea surface oil film with an optimized laser-induced fluorescence spectrum band. Background Art

[0002] As one of the world's primary energy sources, oil often causes marine spills due to accidents or leaks during its extraction, transportation, and utilization, wreaking havoc on marine ecosystems. The resulting oil film on the sea surface blocks sunlight, disrupting photosynthesis in marine plants and undermining the foundation of the marine food chain. Furthermore, toxic components in the oil film directly harm the physiological functions of marine organisms, inhibiting their growth and reproduction. Therefore, rapid and accurate detection and quantification of the thickness of oil films on the sea surface is a key prerequisite for effective pollution control and ecological restoration.

[0003] Currently, fluorescence spectroscopy is a mainstream technique for measuring oil film thickness on the sea surface. It excites fluorescent substances in the oil film and analyzes the emission spectral characteristics to achieve thickness inversion. Traditional fluorescence spectrometers typically scan the entire wavelength range (e.g., 200-800nm) point by point to obtain complete spectral data.

[0004] However, the full-range scanning mechanism of traditional fluorescence spectrometers is inefficient, and the actual effective spectral information may be concentrated in only a few characteristic bands. In low-concentration oil film detection, weak fluorescence signals are easily masked by background noise (such as stray light and detector dark current), resulting in a decrease in signal-to-noise ratio and measurement errors. In addition, in multi-component mixed oil spill scenarios, the fluorescence emission spectra of different oils or pollutants may overlap, making it difficult for traditional spectrometers to achieve accurate distinction through wide-spectrum scanning, limiting their applicability in complex environments.

[0005] Therefore, there is an urgent need for a method to measure the sea surface oil film thickness with an optimized laser-induced fluorescence spectral band. Summary of the Invention

[0006] In view of this, the present invention provides a method for measuring the sea surface oil film thickness by optimizing the laser-induced fluorescence spectral band. The fluorescence spectral data of oil film samples of different thicknesses are collected through laser-induced fluorescence technology, and the Ivy-optimized Sparse Partial Least Squares Algorithm (Ivy-SPLS) is combined to perform band optimization and modeling, so as to achieve fast and accurate measurement of the oil film thickness.

[0007] To this end, the present invention provides the following technical solutions:

[0008] A method for measuring sea surface oil film thickness using an optimized laser-induced fluorescence spectral band, comprising:

[0009] The laser is used to emit an excitation beam to the oil film on the sea surface to be measured, thereby exciting the oil film to generate a fluorescent signal;

[0010] The fluorescence signal is received by a spectrum acquisition device to obtain fluorescence spectrum data of the oil film;

[0011] The characteristic bands are selected from the fluorescence spectrum data by using a sparse partial least squares model, and the sea surface oil film thickness is predicted based on the characteristic bands in combination with Lasso regression; and the selection of characteristic bands is optimized by using the ivy algorithm.

[0012] Furthermore, the optimization of the selection of characteristic bands using the Ivy algorithm includes:

[0013] The inverse of the root mean square error between the predicted value and the true value is used as the fitness value of the Ivy algorithm.

[0014] Furthermore, Lasso regression predicts the sea surface oil film thickness based on characteristic bands, including:

[0015] Lasso regression performs regularized fitting on the feature space to predict the oil film thickness:

[0016] X=TP T +E

[0017] y=TQ T +F

[0018]

[0019] Where T is the latent variable matrix, P and Q are loading matrices, E and F are residual matrices, α is the regularization coefficient, and β is the model coefficient; X represents the spectral data, y represents the oil film thickness, and n represents the number of samples.

[0020] Furthermore, the step of receiving the fluorescence signal through a spectrum acquisition device to obtain fluorescence spectrum data of the oil film further includes:

[0021] receiving the fluorescence signal through a spectrum acquisition device to determine original fluorescence spectrum data of the oil film;

[0022] The original fluorescence spectrum data is preprocessed to obtain fluorescence spectrum data of the oil film.

[0023] Furthermore, the preprocessing includes:

[0024] The Savitzky-Golay smoothing algorithm was used to smooth and denoise the original fluorescence spectrum data to obtain smooth data;

[0025] The smoothed data were normalized using Z-score to transform the data into a distribution with a mean of 0 and a standard deviation of 1.

[0026] Furthermore, a laser with a wavelength of 355 nm is used to excite the oil film to be measured to generate a fluorescence signal.

[0027] Advantages and positive effects of the present invention:

[0028] The method of the present invention selects characteristic waves for oil film detection through a sparse partial least squares model, thereby achieving efficient and accurate acquisition of characteristic wavebands for oil film measurement, avoiding full-band acquisition, and improving detection efficiency; and combining the ivy-optimized sparse partial least squares model to further improve the accuracy of characteristic waveband selection, as well as the robustness of the surface model, significantly improving the accuracy and efficiency of oil film thickness measurement. The optimized sparse partial least squares model is applied to actual marine oil spill monitoring, and by collecting fluorescence spectral data in real time and inputting it into the model, the oil film thickness can be quickly calculated. This method can also be integrated into mobile platforms such as drones or buoys to achieve large-scale, real-time marine oil spill monitoring, providing reliable technical support for environmental protection and oil spill emergency response. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0030] Figure 1 Flow chart of the method for measuring sea surface oil film thickness optimized for the laser-induced fluorescence spectroscopy band;

[0031] Figure 2 This is a diagram of the experimental scene for collecting fluorescence spectrum data in Example 2 of the present invention;

[0032] Figure 3 This is a flow chart of band optimization based on the Ivy sparse partial least squares algorithm in Example 2 of the present invention;

[0033] Figure 4 This is a diagram of the experimental results in Example 3 of the present invention. DETAILED DESCRIPTION

[0034] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0035] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0036] The present invention provides a method for measuring sea surface oil film thickness by optimizing laser-induced fluorescence spectral bands. Through target-oriented data compression, characteristic bands of fluorescence spectral data are selected to reduce invalid acquisition. The signal-to-noise ratio is improved by preprocessing the collected raw fluorescence spectral data. The Ivy algorithm is used to optimize the selection of characteristic bands and enhance dynamic adaptability to accommodate complex samples.

[0037] Combine Figure 1 The method steps of the present invention include:

[0038] S1, using a laser to emit an excitation beam to the oil film on the sea surface to be measured, to excite the oil film to generate a fluorescent signal;

[0039] S2. Receive the fluorescence signal through the spectrum acquisition device to obtain the fluorescence spectrum data of the oil film;

[0040] S3. Feature bands are selected from the fluorescence spectrum data through the sparse partial least squares model, and the sea surface oil film thickness is predicted based on the feature bands in combination with Lasso regression; at the same time, the Ivy algorithm is used to optimize the selection of feature bands.

[0041] Example 1

[0042] The oil film is excited by a 355nm wavelength laser to generate a fluorescence signal. The data quality is improved by optimizing the detection angle and spectrometer parameters.

[0043] The fluorescence signal is smoothed, denoised and standardized to eliminate noise and dimension differences and obtain fluorescence spectrum data;

[0044] Feature band selection and modeling are performed based on ivy optimization combined with sparse partial least squares algorithm: feature bands are extracted from high-dimensional fluorescence spectral data, and the selection of feature bands is optimized by simulating the ivy growth process. The optimal band combination is output for accurate prediction of oil film thickness; and sparse constraints are introduced to enhance the interpretability and generalization ability of the model.

[0045] Example 2

[0046] S1. Construct the original fluorescence spectrum dataset of oil film samples with different thicknesses.

[0047] Oil film samples of different thicknesses were prepared, including: 0# diesel, engine oil CF-4 and 92# gasoline; the oil film thickness covered the range of 0-8mm, which was specifically divided into three intervals: 0-0.08mm (interval 0.008mm), 0.08-0.8mm (interval 0.08mm) and 0.8-8mm (interval 0.8mm).

[0048] use Figure 2 In the experimental system shown, the oil film sample and blank control sample were placed on a precalibrated experimental platform, ensuring that the laser beam could illuminate the oil film sample surface perpendicularly. A Nd:YAG solid-state laser (Nimma-400) was used as the excitation light source, emitting ultraviolet laser light with a wavelength of 355 nm, an operating frequency of 10 Hz, and a pulse energy of up to 100 mJ. The laser illuminated the oil film sample at a 45-degree angle through a plane mirror, stimulating a fluorescence signal. The fluorescence signal was received by a fiber optic probe at a 15-degree tilt angle to reduce interference from the direct excitation light source and total reflected light. The spectrometer used a portable field spectrometer (ASDFieldSpec3) with a measurement range of 350 nm to 2500 nm and an accuracy of ±1 nm. In this example, a detection range of 370 nm to 700 nm was used. Ten fluorescence spectra were collected for each oil film sample thickness. The spectrometer's averaging function was used to reduce random errors, ultimately resulting in 870 raw fluorescence spectra, which served as the raw fluorescence spectral dataset.

[0049] S2, performing smoothing, denoising and standardization preprocessing on the original fluorescence spectrum dataset to eliminate noise and dimensional differences, and obtain the fluorescence spectrum dataset of the oil film;

[0050] The collected raw fluorescence spectral data usually contains noise and baseline drift, so preprocessing is required to improve data quality. First, the Savitzky-Golay smoothing algorithm is used to smooth and denoise the spectral data. This algorithm smoothes the data by fitting a polynomial within a local window, which can effectively remove noise while retaining the characteristic peaks of the spectrum. The smoothed data is further standardized to eliminate the dimension and amplitude differences between different samples. The normalization method uses Z-score normalization to convert the data into a distribution with a mean of 0 and a standard deviation of 1 as a fluorescence spectral dataset; the preprocessed fluorescence spectral dataset not only improves the signal-to-noise ratio, but also provides high-quality input for subsequent band optimization and modeling.

[0051] S3. Feature bands are selected from the fluorescence spectrum data through the sparse partial least squares model, and the sea surface oil film thickness is predicted based on the feature bands in combination with Lasso regression; at the same time, the Ivy algorithm is used to optimize the selection of feature bands.

[0052] Combine Figure 2 In this embodiment, a sparse partial least squares model is used to select characteristic bands from the fluorescence spectrum data, and Lasso regression is combined with the characteristic bands to predict the sea surface oil film thickness. At the same time, the Ivy algorithm is used to optimize the selection of characteristic bands. Further explanation:

[0053] 1) Randomly generate a set of initial members, each member represents a possible band combination or parameter setting. Randomly determine the initial position of the members in the search space:

[0054] I i =I min +rand(1,D)×(I max -I min )

[0055] Among them, I i represents the initial position of the i-th member, i = 1, 2, ..., Npop; Npop represents the total number of members, D represents the number of decision variables, and the D-dimensional vector of random numbers uniformly distributed in the interval [0, 1] is represented by rand(1, D); defines the maximum value of the spectral data I max As the upper bound of the search space, the minimum value of the spectral data I min As the lower bound of the search space; the i-th member is in the form of I i =(I i1 ,...,I iD ), where i = 1, 2, ..., Npop, the initial population of ivy plants is represented by

[0056] 2) Explore a better solution space by simulating the growth and spread of ivy:

[0057] The growth rate of ivy is calculated using the differential equation:

[0058]

[0059] ΔGv i (t+1)=rand 2 ×(N(1,D)×ΔGv i (t))

[0060] Among them, Gv(t) represents the growth rate of members, ψ represents the growth rate, Indicates deviation from the optimal growth path; ΔGv i(t+1) represents the growth rate at time t, and N(1,D) represents a random vector of dimension D.

[0061] 3) The characteristic bands are selected from the fluorescence spectrum data through the sparse partial least squares model, and the sea surface oil film thickness is predicted based on the characteristic bands in combination with Lasso regression:

[0062] Each solution is substituted into the partial least squares model to extract latent variables and reduce data dimensionality. Lasso regression is used to perform regularized fitting on the feature space to predict oil film thickness. Lasso regression introduces an L1 regularization term to constrain model coefficients and reduce redundant variables. The formula is expressed as:

[0063] X=TP T +E

[0064] y=TQ T +F

[0065]

[0066] Where T is the latent variable matrix, P and Q are loading matrices, E and F are residual matrices, α is the regularization coefficient, and β is the model coefficient; X represents the spectral data, y represents the oil film thickness, and n represents the number of samples.

[0067] The root mean square error between the predicted value and the true value is calculated, and the inverse of RMSE is used as the fitness value.

[0068]

[0069] Fitness = 1 / RMSE

[0070] Among them, y i Indicates the actual oil film thickness, represents the predicted oil film thickness, m is the number of samples; RMSE represents the root mean square error between the predicted value and the true value; Fitness represents the fitness value.

[0071] 4) According to the fitness function value, select I i The nearest most important neighbor; and construct a member sequence based on fitness sorting The formula is:

[0072]

[0073] in, Represents a member sequence; I ii Represents the nearest most important neighbor determined by the fitness function value; I best represents the best member of the entire population; Represents the member ranked j in the member sequence.

[0074] Member I i Use its nearest most important neighbor member I ii Climbing and moving logically in the direction of the light source in the equation, the formula is expressed as:

[0075]

[0076] Among them, |N(1,D)| is a vector whose components are the absolute values ​​of the components of vector N(1,D); member I i Follow I directly best , Indicates the new position after climbing. ii Perform a global search to find best Seeking improved solutions nearby.

[0077] 5) Iterative optimization by repeatedly performing 1)-4)

[0078] Each iteration generates a new population and merges it with the previous generation population to form a new population Sort by fitness value from most favorable to least favorable to obtain the member ranking of the new population And select the first Npop members in the sort as the member sort of the next round of population The formula is:

[0079]

[0080] Until the number of iterations I ter Reaching the preset iteration threshold I termax , stop iteration and output the final member;

[0081] 6) The final member is output as the optimal solution of the Ivy sparse partial least squares, which represents the optimal band combination for laser-induced fluorescence spectroscopy measurement of oil film thickness; and the corresponding PLS model parameters are obtained.

[0082] Example 3

[0083] The effectiveness of this method is verified through comparative experiments:

[0084] Oil film thickness detection experiments were conducted on test sets of different oils, and the RMSE interval plots of this method and existing detection methods were compared to perform cross-validation results.

[0085] The prediction methods used in the comparative experiment include:

[0086] partial least squares (PLS, partial least squares regression);

[0087] Variable importance in projection (VIP);

[0088] Backward variable selection for PLS (BVSPLS, backward variable selection based on PLS);

[0089] Genetic algorithm for wavelength selection (GA);

[0090] Wavelength band selection using simulated annealing (SA);

[0091] Moving window partial least squares (MWPLS)

[0092] The experimental results are as follows Figure 4 As shown:

[0093] Figure 4 (a) is the test result of cross-validation of the RMSE interval diagram of different oil film detection methods on the 0# diesel test set;

[0094] Figure 4 (b) shows the test results of the cross-validation results of the interval plot of RMSE of different oil film detection methods on the engine oil CF-4 test set;

[0095] Figure 4 (c) is the test result of cross-validation of the RMSE interval diagram of different oil film detection methods on the 92# test set;

[0096] It can be seen that after using this method to optimize the band, the prediction accuracy of the model is significantly improved, the root mean square error is reduced to below 0.02mm, and the correlation coefficient (R 2 ) can reach as high as 0.996 and above, reducing measurement error by over 30% compared to traditional methods. Compared to traditional methods (such as full-band PLS and simulated annealing algorithms), the proposed method improves computational efficiency by 50% and exhibits strong robustness to noise and interference. Furthermore, this method is highly adaptable to different types of oil film samples, and can accurately measure the thickness of various oil films, including 0# diesel, CF-4 engine oil, and 92# gasoline.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for measuring sea surface oil film thickness using an optimized laser-induced fluorescence spectroscopy band, characterized in that: include: The laser is used to emit an excitation beam to the oil film on the sea surface to be measured, thereby exciting the oil film to generate a fluorescent signal; The fluorescence signal is received by a spectrum acquisition device to obtain fluorescence spectrum data of the oil film; The characteristic bands are selected from the fluorescence spectrum data by using a sparse partial least squares model, and the sea surface oil film thickness is predicted based on the characteristic bands in combination with Lasso regression; and the selection of characteristic bands is optimized by using the ivy algorithm.

2. The method for measuring sea surface oil film thickness using a laser-induced fluorescence spectrum band optimization method according to claim 1, characterized in that: The method of optimizing the selection of characteristic bands using the Ivy algorithm includes: The inverse of the root mean square error between the predicted value and the true value is used as the fitness value of the Ivy algorithm.

3. The method for measuring sea surface oil film thickness using a laser-induced fluorescence spectrum band optimization method according to claim 1, characterized in that: Lasso regression predicts sea surface oil film thickness based on characteristic bands, including: Lasso regression performs regularized fitting on the feature space to predict the oil film thickness: X=TP T +E y=TQ T +F Where T is the latent variable matrix, P and Q are loading matrices, E and F are residual matrices, α is the regularization coefficient, and β is the model coefficient; X represents the spectral data, y represents the oil film thickness, and n represents the number of samples.

4. The method for measuring sea surface oil film thickness using a laser-induced fluorescence spectrum band optimization method according to claim 1, characterized in that: The fluorescence signal is received by a spectrum acquisition device to obtain fluorescence spectrum data of the oil film, and further includes: receiving the fluorescence signal through a spectrum acquisition device to determine original fluorescence spectrum data of the oil film; The original fluorescence spectrum data is preprocessed to obtain fluorescence spectrum data of the oil film.

5. The method for measuring sea surface oil film thickness using a laser-induced fluorescence spectrum band optimization method according to claim 4, characterized in that: The preprocessing comprises: The Savitzky-Golay smoothing algorithm was used to smooth and denoise the original fluorescence spectrum data to obtain smooth data; The smoothed data were normalized using Z-score to transform the data into a distribution with a mean of 0 and a standard deviation of 1.

6. The method for measuring sea surface oil film thickness using a laser-induced fluorescence spectrum band optimization method according to claim 1, characterized in that: The oil film to be measured is excited by laser light with a wavelength of 355 nm to generate a fluorescent signal.

Images