<p>Support Vector Machine (SVM) is a widely used pattern classification method, and its classification accuracy is significantly influenced by the selection of penalty factors and kernel parameters. Swarm intelligence optimization algorithms, as a method for solving optimization problems, have been extensively applied to the parameter optimization of SVM. However, most intelligent optimization algorithms have shortcomings such as easily falling into the local optima or slow convergence speed. Therefore, this paper proposes an improved Chaotic Adaptive Whale Optimization Algorithm (CAWOA) to optimize the parameters of SVM, thereby enhancing the performance of SVM. The improved CAWOA incorporates strategies such as logistic chaotic mapping for population initialization, non-linear convergence factors, and adaptive probability thresholds, based on the standard WOA. Firstly, simulation comparison experiments were conducted on the CEC2017 and CEC2022 test sets, and the results showed that the improved algorithm has better performance. Then, a CAWOA-SVM model was established, and classification predictions were made on six datasets from the UCI database. The CAWOA-SVM model was compared with WOA-SVM, PSO-SVM models, and the non-optimized SVM model. The experimental results indicate that the classification accuracy and overall performance of the CAWOA-SVM model have been improved to some extent.</p>