The increased penetration of wind power generation has introduced significant uncertainty and complexity to power system operations, making the real-time dynamic security assessment (DSA) a necessity to protect the system against the risk of blackouts. As a promising strategy for real-time DSA, the intelligent system (IS)-based approach extracts the DSA knowledge from a dynamic security knowledgebase (KB). The quality of KB is the key to the success of such an IS. The conventional Monte Carlo (MC) method requires a large number of sampled cases to cover the operating point (OP) space due to its random sampling mechanism, and therefore is computationally expensive. To generate a KB more efficiently, this paper proposes a Good Point Set (GPS)-based KB generation scheme where the sample distribution is more uniform. To achieve the same level of DSA accuracy, the GPS-based approach requires less number of samples. A case study is conducted on power system with high wind penetration and its result verifies that the GPS method outperforms conventional MC method.