ABSTRACT The share of renewable energy generation in the power system has been continuously increasing, imposing a significant computational burden on planning models due to the consideration of a large amount of time‐varying data. Existing time‐series aggregation (TSA) methods struggle to achieve an optimal balance between modelling accuracy and solution efficiency. To this end, this paper proposes a novel variable‐scale time series aggregation (VSTSA) method. By breaking the constraints of fixed‐scale clustering, VSTSA employs a multi‐level sliding window mechanism to capture net load fluctuation processes of varying durations. A set of variable‐scale typical scenarios is constructed to cover both long‐period inter‐day fluctuations and short‐period variations, with a specific mechanism to retain extreme periods. Variable‐scale scenarios are applied to planning models to simplify the calculation of annual operational simulations. Each typical period represents a cluster of identical operational states, weighted by their coverage in the original time series. Case studies demonstrate that, compared with existing clustering techniques, VSTSA not only significantly improves computational efficiency in generation capacity planning but also enhances calculation accuracy by retaining extreme periods not covered by typical scenarios, thereby achieving a superior balance between accuracy and efficiency.
Zhang et al. (Thu,) studied this question.