Adaptive comprehensive learning particle swarm optimization (ACLPSO) is a powerful particle swarm optimization variant. In ACLPSO, the interval scaling coefficient s and learning probability tradeoff coefficient v are key parameters dominating convergence. ACLPSO statically sets values for s and v. However, reported experimental results have shown that ACLPSO may require different value pairs of s and v on different benchmark functions to derive the global optimum or a near-optimum with high fitness accuracy, and there is no rule on determining the most appropriate value pair. To eliminate the need of manually tuning s and v, this paper proposes to dynamically set values for the two coefficients. Experimental results on various benchmark functions demonstrate that ACLPSO with dynamic s and v is able to find the global optimum or a near-optimum with high fitness accuracy on most functions, significantly better than ACLPSO with each static value pair of s and v.