基于MLS算法的汽车点火线圈铁芯磁化曲线拟合

针对已有点火线圈建模方法因忽略铁芯磁化过程而导致精度较低的问题,基于局部近似的思想利用移动最小二乘方法对其基本磁化曲线进行拟合.为抑制拟合中的粗差数据干扰问题,利用初次拟合误差判定实测数据中的粗差点,并通过二次拟合修复粗差数据,拟合后的总体平均相对误差低于0.53%.在此基础上,将拟合后的曲线代入到点火线圈的等效磁路-电路计算模型中,计算得到了火花电流、火花电压与火花能量的仿真波形与数值,通过与实际测试的波形对比,验证了拟合算法的有效性和合理性,为点火线圈性能分析和结构设计提供了理论依据.

Magnetization Curve Fitting of the Ignition Coil Core Based on Moving Least Squares

According to the low precision problem caused by neglecting the magnetization process of the ignition coil core in tradition methods,moving least squares(MLS) is applied to fit the magnetization curve based on the local approximation theory,meanwhile,the gross errors of measured data are detected and rejected based on absolute errors threshold of the first fitting,then the repaired data are applied to the second curve fitting.The total average relative errors are less than 0.53%.Furthermore,the fitted curve is substituted into the simulation model of ignition coil,the value and waveforms of ignition current,spark time,and ignition energy are calculated.Comparison of result verifies the validity and rationality of the algorithm,which provides a theoretical basis for performance analysis and structure design of ignition coil.