Abstract: The damage strength and the peak strength are the important indexes for rock engineering. The two indexes identified by the acoustic emission (AE) tests are of high practical value in engineering application. As it is difficult to identify the two indexes by the AE tests, the corresponding identification method is studied based on the laboratory AE tests. Firstly, the AE tests on typical red sandstone samples are carried out, and the whole process of the uniaxial compression tests can be divided into three stages on the subject to damage variables characterized by the number of AE events. The three stages include the stable evolution period of damage (before damage strength), the aggravated evolution period of damage (between damage and peak strengths) and the residual strength period after the peak strength. The appropriate AE parameters for identification are selected based on the Spearman correlation coefficient between AE and damage state. Then the identification model is established based on the principle of SVM classification. It can be used to identify the stages before and after the damage strength and peak strength of rock. The RBF kernel function and the PSO algorithm are determined as the optimal algorithm based on the analysis of different kernel functions and parameter optimization algorithms. The itentification accuracy increases with the decrease of the difference of wave velocity between the test and training samples. With the close wave velocity between the test and training samples, the identification accuracy of the three stages is over 96%. The research results may provide reference for identifying the strength states of in-situ rock through AE monitoring.