Modeling And Fault Diagnosis Of Wind Turbine System Based On Bond Graph

Wind energy is one of the most important clean energy. In order to make full use of wind energy, the developing trend of wind turbine system is large scale, and this leads to higher complexity and failure rate. In order to improve the reliability and reduce the maintenance cost and failure time, the wind turbine system should be effective in fault diagnosis. Wind turbine system is a multi-energy domain complex system, traditional modeling method can’t descript the physical structure well and can’t obtain a unified model, which make the analysis unintuitive and difficult to diagnosis in a unified method. Bond graph is a unified modeling method which can be used to model multi-energy domain complex systems and represent its physical structure. Therefore, the wind turbine system modeling and fault diagnosis based on bond graph are studied in this paper.Firstly, a bond graph diagnosis method based on temporal causal graph(TCG) is proposed in this paper for fault detection and isolation(FDI) of system components in a systematical way. The proposed method used for FDI is based on analytical redundancy relations(ARRs) which derived from junction function of bond graph model. In order to obtain as much ARRs as possible to improve the isolability, first of all, the method obtains TCG from bond graph model; then, according to TCG, obtains relations graph of variables; finally, eliminates unknown variables of junction functions by relations graph of variables. In the process of eliminating unknown variables, the backward and forward propagation of causal paths are used so that more different ARRs can be obtained under the same condition of given sensors placement. This method can be used for all kinds of system because of the multi-energy domain unified modeling feature of bond graph.Secondly, the bond graph models of the blade and pitch system, the drive train system, the converter and generator which composed of wind turbine system are built in this paper. According to these models, ARRs and fault signature matrix(FSM), the basis of fault diagnosis, are derived by the presented method. Moreover, a benchmark model of wind turbine system is presented which can be used to simulate the running state of wind turbine system under actual wind speed.Finally, the models and fault diagnosis method are tested by simulation. The results show the proposed method can be used to the fault diagnosis of the main components of wind turbine system.