Abstract: Focused on the characteristics of time-varying, nonlinear and difficult load forecasting of central air conditioning system,an improved artificial bee colony algorithm is proposed to optimize support vector regression, and a central air-conditioning load forecasting model is established.Firstly,the search strategy of artificial bee colony algorithm is improved to improve the local search ability and development ability.Secondly, the optimized parameters are optimized by using the improved algorithm to optimize the parameters of the support vector regression machine.Finally, theoretical analysis and case validation show that the load forecasting model based on improved support vector regression has higher prediction accuracy and improves the effectiveness and practicability of data.