Setting and generation of power system typical testing scenarios based on mapping deduction

This article proposes the typical testing scenarios setting and generation method for quantitatively assessing the intelligence level of power grid control agents by mapping deduction principle. This method achieves the setting of typical testing scenarios by comprehensively considering the quantitative intelligence assessment method of power grid control agents, the testing scenario requirements of intelligence indexes, and the basic settings of training scenarios of power grid control agents. Subsequently, this article takes the probability distribution model and the Latin hypercube sampling method to generate power grid typical testing scenarios meeting the testing and assessment requirements of the intelligence indexes. This article further conducts three experiments, as the power grid corrective control agent, the power grid thermal stability control agent, and the power grid transient stability control agent. The experimental results show that the typical testing scenario setting and generation method proposed succeeds in generating testing scenarios meeting the requirements of intelligence indexes for the quantitative intelligence assessment. These testing scenarios can be effectively applied to the testing and operation of power grid control agents, as well as their quantitative intelligence assessment and self-evolution. •A typical testing scenarios generation method for power grid control agents.•The method generates typical testing scenarios for intelligence assessment.•Typical scenarios are qualified for agents’ quantitative intelligence assessment.•Typical scenarios could be applied to the PLASE system.