CN-121185874-B - LIBS-based metal particle size estimation method and device and LIBS detection system

CN121185874BCN 121185874 BCN121185874 BCN 121185874BCN-121185874-B

Abstract

The application relates to the technical field of spectrum detection, and discloses a LIBS-based metal particle size estimation method and device and a LIBS detection system, wherein the method comprises the steps of obtaining LIBS spectrum data of samples with different particle sizes; the method comprises the steps of performing feature variable screening on LIBS spectrum data to obtain optimal interval combinations of wavelengths, training the optimal interval combinations to obtain candidate model parameters by taking the optimal interval combinations as input quantities and the metal particle sizes as response quantities, performing k-fold cross validation on candidate latent variable numbers by means of MC-PLSR modeling, and combining the candidate model parameters to obtain optimal latent variable numbers, wherein the latent variable numbers represent relations between spectrum information and particle sizes, and retraining an MC-PLSR model by means of the optimal latent variable numbers and the LIBS spectrum data to obtain a particle size-LIBS signal mapping function constructed by a final weight vector and a final regression coefficient vector. The dependence of the traditional particle size characterization on high-resolution imaging equipment is eliminated, and the high-precision and reliable prediction of the particle size of the metal particles in the lubricating oil is realized under the condition of no intervention of a precise imager.

Inventors

LU YUAN
LI NING
GUO JINJIA

Assignees

中国海洋大学

Dates

Publication Date: 20260508
Application Date: 20251015

Claims (7)

1. A LIBS-based metal particle size estimation method, comprising: s1, acquiring LIBS spectrum data of samples with different metal particle sizes; S2, screening the LIBS spectrum data to obtain an optimal interval combination of the wavelengths of each sample, wherein the LIBS spectrum data is subjected to characteristic variable screening to obtain the optimal interval combination of the wavelengths of each sample, and the method comprises the steps of dividing the LIBS spectrum data into a plurality of continuous wavelength intervals, optimizing different combinations of each wavelength interval and each wavelength interval, and outputting a physical wavelength range of the optimal interval combination by taking minimization of cross validation errors and maximization of decision coefficients as decision basis; S3, training the optimal interval combination serving as an input quantity and the metal particle size serving as a response quantity to obtain candidate model parameters, wherein the training the optimal interval combination serving as the input quantity and the metal particle size serving as the response quantity to obtain the candidate model parameters comprises the steps of carrying out mean value centering on the input quantity and the response quantity to obtain a centering matrix, constructing a constraint optimization problem by utilizing the centering matrix, applying monotonicity constraint, and iteratively updating a weight vector and a regression coefficient vector of a model; S4, determining the latent variable number, namely performing k-fold cross validation on candidate latent variable numbers by using an MC-PLSR model, and combining the candidate model parameters to obtain optimal latent variable numbers, wherein the latent variable numbers represent the relation between spectral data and metal particle sizes, the k-fold cross validation is performed on the candidate latent variable numbers by using the MC-PLSR model, and the optimal latent variable numbers are obtained by combining the candidate model parameters, and the method comprises the steps of dividing the LIBS spectral data into a training set and a validation set, performing MC-PLSR model modeling on each fold, applying monotonic constraint, performing k-fold cross validation on the candidate latent variable numbers to obtain cross validation root mean square error corresponding to each candidate latent variable number, and taking the candidate latent variable number corresponding to the average value of the minimum cross validation root mean square error as the optimal latent variable number; s5, a final training step of retraining an MC-PLSR model by utilizing the optimal latent variable number and the LIBS spectrum data to obtain a 'particle size-LIBS signal' mapping function constructed by a final weight vector and a final regression coefficient vector.
2. The LIBS-based metal particle size estimation method according to claim 1 wherein constructing constraint optimization problems and imposing monotonicity constraints using the centralization matrix comprises: Constructing a constraint optimization problem of an NIPALS algorithm by using the centralization matrix, and iteratively updating a weight vector and a regression coefficient vector of a model, wherein the constraint optimization problem comprises an objective function, an optimization target and constraint conditions; and after updating the weight vector in each iteration, monotonicity constraint is applied, and regression coefficient vectors corresponding to the model are projected to a non-negative orthogonal space.
3. The LIBS-based metal particle size estimation method according to claim 1 wherein the iteratively updating the weight vector and regression coefficient vector of the model comprises: calculating a score vector according to the iteratively updated weight vector; regression is carried out on the input quantity and the output quantity by utilizing the score vector, and a residual matrix is calculated; and iteratively calculating the score vector and the residual matrix until a preset convergence condition is met, and ending the iteration.
4. A LIBS-based metal particle size estimation method according to any one of claims 1 to 3 wherein after the acquisition of LIBS spectral data for samples of different particle sizes the method further comprises: Constructing a sample intensity matrix by using the LIBS spectrum data; analyzing the sample intensity matrix by using a PCA algorithm, and extracting key characteristic spectral lines for distinguishing metal particle types; and constructing a PCA metal classification model according to the key characteristic spectral lines.
5. A LIBS-based metal particle size estimation apparatus comprising a processor and a memory storing program instructions, wherein the processor is configured to perform the LIBS-based metal particle size estimation method of any one of claims 1 to 4 when the program instructions are run.
6. A LIBS detection system comprising: a laser for providing laser light required for excitation of the sample; The optical path system is used for receiving and adjusting the laser output by the laser to enable the laser to be focused on the surface of the sample and excite the sample to generate plasma; the spectrum acquisition system is used for receiving the emitted light of the plasmas output by the light path system and analyzing the emitted light to obtain spectrum intensity; The LIBS-based metal particle size estimation apparatus according to claim 5 in communication with the spectrum acquisition system.
7. A computer-readable storage medium storing program instructions that, when executed, cause a computer to perform the LIBS-based metal particle size estimation method according to any one of claims 1 to 4.

Description

LIBS-based metal particle size estimation method and device and LIBS detection system Technical Field The application relates to the technical field of spectrum detection, in particular to a LIBS (laser induced breakdown spectroscopy) -based metal particle size estimation method and device and a LIBS detection system. Background Lubricating oils play a critical role in protecting and maintaining operation of the engine. Under high temperature and high pressure operating conditions, friction inevitably occurs between components within the engine, and micron-sized metal wear particles (e.g., al, fe, cu, cr, etc.) are generated and incorporated into the lubricating oil. Over time, severely contaminated lubricating oils become progressively blackened, which is a typical manifestation of oil aging and the suspension of large amounts of metal particles. Particle size identification of metal particles in lubricating oils currently relies primarily on microscopic imaging techniques, i.e., estimating the actual particle size by analyzing the number of pixels of the particle image. However, this method is severely dependent on image quality. When the lubricating oil is blackened due to pollution, the imaging contrast is reduced, the grain boundary is blurred, and the grain diameter is easy to misjudge. At this time, the oil sample needs to be repeatedly cleaned to remove the interference of the ground color, so that accurate measurement can be realized. However, this process is tedious and time-consuming, and is difficult to meet the requirements of rapid analysis and batch statistics. It should be noted that the information disclosed in the above background section is only for enhancing understanding of the background of the application and thus may include information that does not form the prior art that is already known to those of ordinary skill in the art. Disclosure of Invention The following presents a simplified summary in order to provide a basic understanding of some aspects of the disclosed embodiments. This summary is not an extensive overview, and is intended to neither identify key/critical elements nor delineate the scope of such embodiments, but is intended as a prelude to the more detailed description that follows. The embodiment of the disclosure provides a LIBS-based metal particle size estimation method and device and a LIBS detection system, so as to realize reliable analysis of the metal particle size in contaminated oil. In some embodiments, the LIBS-based metal particle size estimation method comprises a data acquisition step of acquiring LIBS spectrum data of samples with different particle sizes, a screening step of screening feature variables of the LIBS spectrum data to obtain optimal interval combinations of wavelengths of each sample, an initial training step of training the optimal interval combinations as input quantities and the metal particle sizes as response quantities to obtain candidate model parameters, a latent variable number determination step of performing k-fold cross validation on candidate latent variable numbers by MC-PLSR modeling and combining the candidate model parameters to obtain optimal latent variable numbers, wherein the latent variables represent relations between spectrum information and particle sizes, and a final training step of retraining an MC-PLSR model by using the optimal latent variable numbers and the LIBS spectrum data to obtain a particle size-LIBS signal mapping function constructed by a final weight vector and a final regression coefficient vector. In some embodiments, the LIBS-based metal particle size estimation apparatus includes a processor and a memory storing program instructions, the processor being configured to perform the LIBS-based metal particle size estimation method described previously when the program instructions are executed. In some embodiments, the LIBS detection system comprises a laser, an optical path system, a spectrum acquisition system and a metal particle size estimation device based on LIBS, wherein the laser is used for providing high-energy pulses required by excitation of a sample, the optical path system is used for receiving and adjusting laser output by the laser, enabling the laser to be focused on the surface of the sample and exciting the sample to generate plasma, the spectrum acquisition system is used for receiving emitted light of the plasma output by the optical path system and analyzing the emitted light to obtain spectrum intensity, and the metal particle size estimation device based on LIBS is in communication connection with the spectrum acquisition system. In some embodiments, the computer readable storage medium stores program instructions that, when executed, perform the aforementioned method of LIBS-based metal particle size estimation. The LIBS-based metal particle size estimation method and device and the LIBS detection system provided by the embodiment of the disclosure can realize the followi