基于贝叶斯置信框架的最小二乘支持向量机模型参数选择方法及SS_7型电力机车牵引电机建模

Selection Method for Model Parameters Based on Least Squares Support Vector Machine Using Bayesian Evidence Framework and the Modeling of SS_7 Traction Motor of Electric Locomotive

利用贝叶斯置信框架推断LS—SVM模型参数并建立贝叶斯LS—SVM曲线拟合模型。运用贝叶斯LS—SVM曲线拟合模型对SS7型电力机车牵引电机的磁化曲线进行拟合。结果表明:贝叶斯LS—SVM曲线拟合模型对SS7型电力机车牵引电机的磁化曲线拟合速度及精度均有较大地提高,小样本条件下的泛化能力更好。依据贝叶斯LS—SVM曲线拟合模型建立了SS7型电力机车牵引电机的仿真模型。仿真计算表明,所建的SS7型电力机车牵引电机仿真模型与机车牵引电机实际运行状况吻合,可用于对SS7型电力机车主电路性能及控制策略的研究。

Bayesian evidence framework is proposed to estimate the LS--SVM model parameters and a Bayesian LS--SVM curve fitting model is built as well. Using Bayesian LS--SVM curve fitting model, the magnetization curve of SS7 traction motor of electric locomotive can be fitted. Results show that Bayesian LS--SVM curve fitting model can obviously improve the fitting precision and velocity for the magnetization curve of SS7 traction motor of electric locomotive. It performs even better in the generalization capability under small data set. Simulation model of SS7 traction motor of electric locomotive is built based on Bayesian LV-SVM curve fitting model. The simulation results show that the proposed model of SS7 traction motor tallies well with the actual operation status of locomotive traction motor, and can be used to study the performance and control strategy of the main circuit of SS7 electric locomotive