Search

The output torque and rotational speed of vehicle motor will change frequently under the unstable conditions of dynamic change of load and speed, which is more likely to lead to system failure and more difficult to diagnose. Therefore, a simulation model and test platform for multi-fault fast switching of permanent magnet synchronous motor were established in this paper, and the operating data of permanent magnet synchronous motor under different faults were obtained. Based on the analysis of common fault diagnosis techniques and development trends, a hierarchical diagnosis scheme was proposed, which combines data layer, feature layer and decision layer. The fault diagnosis strategy of permanent magnet synchronous motor under variable load and speed conditions were studied. The validity and correctness of the fault simulation model and the hierarchical diagnosis strategy were verified by comparing the experimental results with the simulation analysis. The results of fault identification had acceptable accuracy and reached the expected goal of the research.

The estimation of battery state of charge is the key technology and difficulty in the battery management system. model-based extended Kalman filtering (EKF) and H filtering (HIFF) have received attention for better estimation performance in recent years. In this paper, Panasonic NCR 18650GA battery was selected as the research object. first of all, the Thevenin model was established in the MATLAB / Simulink. After that, the bias compensated recursive least squares method was proposed by performing the deviation compensation on the basis of the recursive least squares method, which works well for parameter identification of data with colored noise. Then, an adaptive mixed EKF/HIFF algorithm was designed to estimate the SOC, combining the accuracy of EKF with the robustness of HIFF. And then the joint estimation of model parameters and battery status was realized. Finally, the robustness and accuracy of the proposed joint estimator has been verified by adding different colored noise to the experimental data. The result indicates that the estimation errors of voltage and SOC are less than 0.5% even if add a large colored noise to the experimental data, which makes the SOC estimation more accurate and reliable for the electric vehicles application.