Multi-objective dynamic flexible job shop scheduling (MO-DFJSS) presents an intricate task of creating optimal job schedules in a manufacturing environment characterised by uncertainty and flexibility, while simultaneously balancing multiple, often conflicting objectives. Current approaches integrate genetic programming (GP) with Pareto dominance-based and scalarising function-based multi-objective methods to learn Pareto fronts of scheduling heuristics for MO-DFJSS. However, these approaches often rely on approximate performance indicators, which can complicate achieving the final goal. In contrast, indicator-based multi-objective methods offer a more straightforward way by using performance indicators as the assessment criterion. Despite their effectiveness in other domains, no indicator-based multi-objective methods have been combined with GP for MO-DFJSS to date. Addressing this gap, this paper proposes SMS-MOGP, a fusion of GP with the SMS-EMOA, which is a popular indicator-based multi-objective algorithm, to learn scheduling heuristics for MO-DFJSS. Experiment results demonstrate that SMS-MOGP achieves comparable performance to NSGPII and significantly outperforms MOGP/D in solving the MO-DFJSS problems.