基于知识映射生成式对抗网络的远动点表智能构建技术研究OA

With the development of smart grids,the complexity of substation monitoring systems has significantly increased,making traditional manual point tables compilation methods increasingly insufficient to meet engineering requirements.A knowledge-mapping-based generative adversarial network(GAN)method is proposed for the automatic generation of remote telemetry point tables,aiming to address the dynamic adaptability and compliance issues in the generation of remote telemetry point tables within substation monitoring systems.The proposed method consists of three main components:First,the DeepSets framework is employed to process the variable-dimensional signal sets in substations,extracting global features through permutation-invariant aggregation.Second,a knowledge template library is constructed to encompass multiple dispatching agencies and voltage levels.The TransH algorithm is used to embed dispatching specifications as low-dimensional vectors,explicitly modeling the complex relationships among substation intervals,signals,and dispatching specifications.Finally,a generative adversarial network is designed.Its generator employs knowledge-guided dynamic filtering to produce candidate point tables,while the discriminator ensures generation quality through a multi-scale verification mechanism.Experimental results demonstrate that the proposed method achieves a compliance rate of 99.6%and a constraint satisfaction rate of 98.3%for generated point tables,representing improvements of 5.4%and 16.75%,respectively,compared to traditional methods.