To address the industrial application requirements for rapid detection using Laser-Induced Breakdown Spectroscopy (LIBS), a swift classification method for heat-resistant steel utilizing portable LIBS equipment has been developed. The study involved six types of typical steel as experimental specimens. The classification models were respectively build based on the support vector machine (SVM), recursive feature elimination method combined with support vector machine (RFE-SVM) and linear discriminant analysis combined with support vector machine (LDA-SVM). The prediction accuracy and running time were used as the main evaluation indexes to evaluate each model, and the impact of input data and spectral feature dimension change on model performance was discussed. The results showed that the classification performance of the LDA-SVM model achieved the best performance, the accuracy of classification prediction was 95.28%, and the average time spent on classification was around 0.011 s. This research offers a methodological guide for the rapid classification of steel types using LIBS technology and lays the groundwork for the development of intelligent, rapid detection capabilities for portable LIBS devices.