Research on a LIBS-based detection method of medium-and-low alloy steel hardness

Abstract

Objective: medium and low alloy steel have good mechanical properties and are widely used in the manufacture of key parts such as high-speed rail, automobiles, and aircrafts. Hardness is an important physical property of metal materials, which directly affects the mechanical properties of metal materials as key components. Therefore, the detection of hardness is of great significance. The traditional hardness testing method is to apply force to the sample to produce indentation and then determine the hardness of the sample by the depth of the indentation. Due to the existence of indentation in this method, the sample will be destroyed, so it can only be used for the random inspection of sample hardness parameters, which cannot meet the reality of the hardness testing of the key parts of the equipment in a running state. Therefore, it is necessary to explore a non-destructive testing method for the hardness of low and medium alloy steel based on LIBS testing technology. Method: based on laser-introduced breakdown spectroscopy (LIBS) technology and combined with chemometrics, this study took the medium-and-low alloy steel as the research object to monitor the quality of train wheels. Firstly, LIBS spectra of 12 standard steel samples with different hardness characteristics were acquired and compared. Secondly, the original spectrum was preprocessed using methods such as smoothing correction, multiplicative scatter correction, AsLS, and AirPLS to select the optimal preprocessing method. It was found that the model established by means of multiplicative scatter correction (MSC) combined with baseline estimation and de-noising using sparsity (BEADS) showed the best performance, in which the cross-validation of correlation coefficient (R_cv) was 0.9909 and the standard deviation of cross-validation (RMSECV) was 14.6935. To simplify the calculation and improve model accuracy, the uninformative variable elimination (UVE), competitive adaptive reweighted sampling (CARS), and successive projections algorithm (SPA) were used to select the effective spectral bands and create the partial least squares and partial minimum support vector machine (SVM) models, respectively. The results showed that the CARS-LS-SVM model obtained more accurate prediction results than the CARS-PLS model, whose R_p and RMSEP values were 0.9966 and 9.4800, respectively. Conclusion: it is concluded that the wavelength selection can not only eliminate the interference information, and simplify the modeling and calculation process of LIBS spectral data but also improve the accuracy of the detection model, thus making it feasible to rapidly detect steel hardness. The results show that it is feasible to predict the hardness of the medium-and-low alloy steel using LIBS combined with the LS-SVM model, which can be further used as a technical method to detect the quality of steel and the performance of the key components for the quality supervision of steel.