基于数据-模型混合驱动的电力系统机电暂态快速仿真方法OA北大核心CSTPCD
A Fast Electromechanical Transient Simulation Algorithm for Power System Based on Data and Physics Driven Model
数据驱动建模方法改变了发电机传统的建模范式,导致传统的机电暂态时域仿真方法无法直接应用于新范式下的电力系统.为此,该文提出一种基于数据-模型混合驱动的机电暂态时域仿真(data and physics driven time domain simulation,DPD-TDS)算法.算法中发电机状态变量与节点注入电流通过数据驱动模型推理计算,并通过网络方程完成节点电压计算,两者交替求解完成仿真.算法提出一种混合驱动范式下的网络代数方程组预处理方法,用以改善仿真的收敛性;算法设计一种中央处理器单元-神经网络处理器单元(central processing unit-neural network processing unit,CPU-NPU)异构计算框架以加速仿真,CPU进行机理模型的微分代数方程求解;NPU 作协处理器完成数据驱动模型的前向推理.最后在IEEE-39 和Polish-2383 系统中将部分或全部发电机替换为数据驱动模型进行验证,仿真结果表明,所提出的仿真算法收敛性好,计算速度快,结果准确.
Data-driven modeling has changed the traditional modeling paradigm of generators,which makes traditional electromechanical transient time domain simulation methods fail to be directly applied to power system with new paradigm.Thus,an integrating data-and physics-driven time domain simulation(DPD-TDS)algorithm for electromechanical transient simulation is proposed.The state variables and nodal injection currents are calculated through data-driven model,and network equations are used to calculate nodal voltages.And a preprocessing matrix calculation method for convergence of DPD-TDS improvement is proposed.A central processing unit-neural network processing unit(CPU-NPU)heterogeneous computing architecture is designed to speed up simulation.Differential algebraic equations are solved in CPU and the forward inference of data-driven model is executed in NPU.In IEEE-39 and Polish-2383 systems,some or all generators are replaced by data-driven models for verification.The results show that the convergence,accuracy and calculation speed of the proposed algorithm are exceptionally impressive..
王鑫;杨珂;黄文琦;马云飞;耿光超;江全元
浙江大学电气工程学院,浙江省 杭州市 310027南方电网数字电网集团有限公司,广东省 广州市 510700
动力与电气工程
机电暂态时域仿真数据-模型混合驱动收敛性CPU-NPU异构运算
electromechanical transienttime-domain simulationdata and physics drivenconvergencecentral processing unit-neural network processing unit(CPU-NPU)heterogeneous computing
《中国电机工程学报》 2024 (008)
2955-2964,中插2 / 11
南方电网数字电网集团有限公司科技项目(670000KK52210032);国家自然科学基金项目(51977188). Research Project of China Southern Power Grid Digital Power Grid Group Co.,Ltd(6700KK52210032);Project Supported by National Natural Science Foundation of China(51977188).
评论