Active Demand (AD) of active distribution network changes user's typical power consumption behavior through market guidance mechanism, the load characteristics of power grid change accordingly, which reduces the accuracy of traditional load forecasting technology. To counter the influence of AD on load forecasting for active distribution network, this paper extracts the variables that affect the AD response load as the exogenous input of load forecasting by considering user's response feature, establishes the complete load forecasting model containing AD input under black box framework based on support vector machine optimized by particle swarm, and proposes a new load forecasting method for active distribution network considering AD. Based on practical considerations about consumers’ behavior, a time-varying AD model is built to produce a data set of load including AD effect to test the proposed load forecasting method. The experimental results show that the proposed load forecasting model is more accurate than the load forecasting model without AD input. This work is supported by National Natural Science Foundation of China (No. 51477090).