基于动态模糊神经网络的开关磁阻电机电感与磁链特性建模

A New Inductance and Flux Linkage Modeling Method for Switched Reluctance Motor Based on Dynamic Fuzzy Neural Network

提出了将动态模糊神经网络用于开关磁阻电机建模的新方法,根据试验采样获得的数据建立电感和磁链特性模型,并将该模型应用到整个系统中。与其他建模方法比较,采用动态模糊神经网络获得的电感和磁链模型可在线训练,并具有紧凑的系统结构和强大的泛化能力。建模所得系统仿真同实际系统比较,电流波形基本一致,验证了新的建模方法的正确性和可行性。同时,该建模方法还可以进一步应用到开关磁阻电机的实时控制系统中,为工程设计和调试提供依据。

A new modeling method using dynamic fuzzy neural network （D-FNN） for switched reluctance motor was proposed. It was based on the inductance and flux linkage characterstics, namely experimentally measured data as well as sample data. Compared with other modeling methods, the inductance and flux linkage models based on D- FNN could be trained on line and had the advantages of compact system structure and strong generalization ability. The switched reluctance drive system was simulated with the trained inductance and flux linkage models. In comparison with the actual system, current waves demonstrated similarity. As a result, the correctness and feasibility of the new modeling method were proved. At the same time, the new modeling method could also be applied to actual real time control system to provide the basis for engineering design and debugging.