Abstract: Aiming at the phenomenon of premature convergence and easy to fall into local optimum when the particle swarm algorithm is used to solve reactive power optimization problems, an adaptive chaotic particle swarm algorithm based on Cauchy mutation is proposed. Based on the introduction of adaptive adjustment strategies and chaotic search for the best particles, Cauchy mutation operation is led in when the algorithm falls into a premature convergence state, and performs Cauchy perturbation on the best particles with fitness values ranked in the top 20% to ensure the diversity of particle swarms, thus the ability of the algorithm to jump out of the local optimal solution is enhanced effectively in the later stage. Taking the active power loss as the objective function and incorporating the penalty function terms of the voltage and reactive power output constraints, the simulation calculations of IEEE 14 and IEEE 30 node standard examples have verified the correctness and feasibility of the algorithm.