The integration of renewable energy sources (RESs) into active distribution networks (ADNs) is essential for reducing carbon emissions but often results in voltage fluctuations and violations. This paper proposes a hierarchical voltage control framework that effectively coordinates diverse controllable devices with various response times in an ADN. The framework comprises three stages: day-ahead scheduling of on-load tap changer (OLTC), intra-day optimization for droop slopes and references for droop controllers in Soft Open Points (SOPs) and distributed generators (DGs), and real-time local voltage regulation. Unlike existing approaches, the proposed approach analytically establishes voltage stability constraints and incorporates them into droop slope optimization for local controllers, mitigating voltage oscillation risks. Additionally, a novel deviation-aware optimization method is developed to calculate optimal voltage references. This method treats the deviations between fixed-point voltages and their references as uncertainties and accounts for their impacts on voltage security through chance-constrained programming. Simulation results demonstrate the effectiveness of the proposed framework in improving voltage regulation performance with guaranteed stability.