基于改进双向长短期记忆神经网络的新型电力系统源 网暂态交互模型构建方法研究

Under the background of new power system, the electricity electronization, physical and digital fusion and complex interconnection of the power grid make the transient states change in many aspects such as information, mechanism, simulation, analysis and control in different degree. Taking transient states as the entry point, this paper focuses on the equivalent construction of source-network station model. A new intelligent construction method of source-network interaction model in power system is proposed. Firstly, it systematically explains the technical thoughts and researching plan of modeling through the framework. It explains the core of the intelligent modeling method from transient interaction response, processing transient equivalent model and adaptability research respectively. Secondly, it is proposed to establish the mapping fusion analysis method which combines surface layer with ground layer for sample acquisition. The appropriate intelligent algorithm is selected according to the modeling requirements and sample size. Finally, the WSCC 9-node system with wind farm station is taken as the example to construct the model and verify the proposed method. Through the comparison of images and the evaluation indicators, the model effects are ideal and adaptable, which confirms the rationality of the proposed idea and the effectiveness of the selected algorithm.