<p>The determination of rock strength parameters is a key issue in the design and construction of mountain tunnels. Traditional exploration methods or indoor experiments involve complex processes and provide limited data. Moreover, empirical formulas based on a single indicator fail to match the characteristics of different rock types, making it challenging to meet the requirements of rapid on-site construction. This study conducted on-site point load tests and rebound tests on typical sections of mountain tunnels with two different rock types, and proposed a rapid determination method for surrounding rock strength that integrates multiple on-site tests. Besides, using artificial intelligence technology, random forest models and least squares support vector machine models based on the bitterling fish optimization algorithm were established for sedimentary rocks and metamorphic rocks, respectively, thus achieving fast and efficient acquisition of rock strength parameters. Finally, data analysis of actual engineering projects proves that this method can effectively extract the variation patterns of rock parameters in different sections, and significantly reduce the time and cost of engineering decision-making, which has a significant impact on the safe construction of tunnel excavation.</p>