The electric spring (ES) is an emerging power electronic apparatus that offers flexible and rapid support for voltage regulation and power loss reduction. It outperforms conventional mechanical Var devices with faster response and continuous adjustment. This paper introduces a two-timescale coordination strategy that leverages ESs for Volt-Var control (VVC) in active distribution networks (ADNs). In this model, ESs and legacy devices work together across different timescales to address voltage deviations: legacy devices mitigate slow voltage variations on a long timescale, while ESs provide rapid Var support for stochastic voltage fluctuations on a short timescale. The coordination between these timescales is realized through a two-stage stochastic programming model. Concurrently, to improve computational efficiency, a multi-layer perceptron (MLP) is established to replicate the original nonlinear ES constraint. Consequently, by linearizing MLP, the original non-convex mixed-integer nonlinear optimization model is transformed into a mixed-integer second-order cone programming model, which can be efficiently solved using readily available solvers. Case studies are conducted on the IEEE 33-bus test system. The results demonstrate the effectiveness and benefits of ESs in alleviating voltage violations and enhancing ADN operational efficiency.