摘要
针对观测数据的非等间隔性及波动性严重影响参数辨识精度问题,基于拉格朗日插值公式,推导出了变步长Simpson数值积分公式,并结合灰色系统理论,提出了一种模型参数灰色辨识的改进方法。用此方法对Volterra模型中的参数进行了辨识仿真,与以往的曲线拟合方法进行了对比分析,并简要分析了模型的初值问题。计算研究表明,基于变步长Simpson数值积分公式的灰色辨识方法在处理非等时间间隔以及数据波动性较大的参数辨识问题时稳定性较好,可满足高精度辨识模型参数的要求。
Based on Lagrange interpolation formula,Simpson formula with different steps was deduced to deal with the influence of unequal interval and most fluctuating data on parameter estimation with high precision.According to grey system theory,an improved method of parameter grey estimation was put forward.Parameters of Volterra model were estimated and simulated by this method,and compared with the results of former curve fitting methods.The issue of initial values was also analyzed.The analyses indicate that the improved method with Simpson formula is superior to the former ones when the data observed are unequal interval and most fluctuating,and parameter estimation can be fulfilled with high precision.
出处
《解放军理工大学学报（自然科学版）》
EI
2008年第1期67-71,共5页
Journal of PLA University of Science and Technology(Natural Science Edition)
