According to the characteristics of the steelmaking-continuous production, this paper proposes a two-stage robust optimization(TSRO) model considering the uncertainties of processing time and transportation time, where the sequencing and assignment variables are determined in the first stage, and the timing variables are specified in the second stage. Focusing on the complexity and nonlinearity of the TSRO problem, this paper applies the linear duality theory to transform it into a network optimization problem in the worst-case scenario. To solve the simplified network optimization problem, this paper proposes an evolutionary solution algorithm named covariance matrix adaptation evolution strategy (CMA-ES) and introduces a bottleneck cast-based restart strategy to improve the algorithmic efficiency. Finally, this study carries out various experiments based on randomly synthetic instances. The computational and statistical results show the effectiveness of the proposed scheduling model under uncertainty and the competitiveness of the improved CMA-ES algorithm.