<p>With the rapid advancement of business process management, event logs from process-aware information systems have become increasingly complex. Missing content due to manual errors or system limitations is inevitable, significantly impacting the reliability of business process analysis. Existing research has mainly focused on repairing logs by detecting deviations from models; however, these methods often address only the control-flow issues or individual missing activities and fail to repair multiple deviations within a single trace while neglecting data constraints. Aiming at these problems, this paper proposes a probabilistic optimization-based event-correlated log repair method using data constraints. Firstly, the deviating traces of the log are clustered with the compliance traces. The compliance traces are then filtered based on the clustering results to serve as reference trace clusters for each deviating trace. Secondly, deviating traces are viewed as a set of incorrectly correlated events. The Particle Swarm Optimization algorithm combines the optimal sequence of these events—i.e., the repair results—based on alignment costs and violations of data constraints. Finally, experiments on synthetic and real logs evaluate the proposed method, demonstrating its effectiveness in repairing deviating traces into compliant traces while ensuring data constraints are met.</p>