Aimed at the optimal chiller loading problem in parallel chillers systems, a multi-strategy-based improved sparrow search algorithm is proposed. The aim of the optimization problem is to minimize chillers power consumption, and the partial load ratio of each chiller is used as the optimization variable. In the improved algorithm, firstly, the chaotic sequence mechanism is introduced to improve the quality and diversity of the initial solution. Secondly, in order to enhance the optimization accuracy, the speed concept in the particle swarm algorithm is proposed to update producer position. To avoid the algorithm from falling into a local optimum for a long time, the following strategy of the wolf pack algorithm is combined to update scrounger position and adjust the individual weight adaptively to improve the convergence speed of the algorithm. Finally, two test cases are selected to confirm the performance of the proposed algorithm in detail, and compared with other commonly used algorithms, the improved sparrow search algorithm can save up by 17.8% and 23.97%, respectively. By using the actual system simulation platform, it is verified that the improved algorithm has the advantages of fast convergence, short running time and good robustness.