Multi-Energy Domain Power Bond Graph Modeling And Fault Diagnosis Of Wind Power Systems

Wind energy is a renewable energy.Since the wind power system can convert the power of the wind into energy,its power generation method is very environmentally friendly.Therefore,more and more countries pay attention to it.Wind power generation accounts for an increasing proportion of our country’s power generation.The complex structure and harsh working environment of wind power generation systems mean that a fault in one part of the system will affect the operation of the system,and a serious fault may even lead to system downtime or damage resulting in large economic losses.Therefore,by adopting fault detection and identification methods,detecting faults at an early stage can effectively stop the propagation of faults and reduce economic losses.The wind power system is a large,complex system with coupled,non-linear multi-energy domains.Power bond graph is a multi-energy domain modeling method,which can not only build a complex multi-energy domain system model with a unified method,but also use a unified method to analyze and diagnose the model.Firstly,the basics of bond graph theory are briefly described.For wind power systems in multiple energy domains,models of the wind energy conversion system,the drive train,the inverter,the Boost booster system and the generator system are built from the perspective of energy traditions and conversions based on the bond graph theory.These system models are coupled into a power hybrid bond graph model of the wind power system.The simulation results prove the reliability and correctness of the model.Secondly,based on the power bond graph model of wind power systems,a diagnostic hybrid bond graph fault diagnosis method based on global analytical redundancy relation is proposed.The hybrid bond graph model of the system is transformed into a diagnostic hybrid bond graph model,the global analytical redundancy relation equation is derived from the characteristic equations of the model nodes and the causal relationships of the components,the global analytical redundancy relation equation is analyzed about the components to obtain the fault feature matrix,and the fault is detected and isolated by matching the residual vector and the fault feature vector.Aiming at the interference of parameter uncertainty on the diagnosis results,a threshold setting method based on the combination of linear fractional transformation and system model is proposed.Based on the linear fractional transformation method and the SIDOPS+ description language,the system model is transformed and simplified,and a new global analytical redundancy relation is derived to construct the adaptive threshold of the system,and the residual is compared with the threshold to determine the residual vector.Finally,the fault diagnosis method is verified using the wind power model on the20-sim simulation platform.The results show that the method proposed in this paper can detect faults in wind power systems quickly and accurately.