Distribution network systems usually contain static and dynamic volt-ampere-reactive (VAR) sources such as photovoltaics (PVs) and static VAR compensators (SVCs). It is therefore necessary to optimize the reactive power while considering the flow and voltage constraints in a distribution network. In this study, we aim to minimize the reactive power appearing in the distribution network due to integrated PVs, encompassing devices such as SVCs and capacitors. On the basis of this principle, first, the restriction of switching times of the capacitor bank under the constraint condition is converted into the adjustment cost of the capacitor bank in the objective function. The coupling variables of the capacitor bank can therefore be decoupled. Second, a multi-objective improved Salp Swarm Algorithm (ISSA) for static and dynamic reactive powers is introduced with the involvement of quasi-reverse learning, chaotic mapping, dynamic inertia weight, and a differential mutation operator. The performance fitness function is then established to determine the best solution for minimizing reactive power. The performance results show that the total network loss of static reactive power due to optimization decreases by 23.25% and the maximum network loss decreases by 20.21%. We also verified that the voltage per unit (pu) values meet the criterion of 0.95 pu with a maximum voltage rise of 3.23%. Dynamic reactive power with optimization presents similar network loss and voltage levels, while reducing the number of operations for each capacitor bank by 75%.