摘要
采用基于最小二乘法的多项式拟合与改进的BP神经网络法分别对同一水泵性能参数进行拟合,并通过算例比较分析了两者拟合效果、精度。结果表明,改进的BP神经网络拟合曲线效果和精度较高,在样本数量不多情况下两者拟合曲线效果较接近。
Polynomial fitting based least square method and improved BP neural network are use to identify the parameters of pump performance curves.The fitting effect and precision of two methods are comparative analysis with examples.The results show that improved BP method has good fitting effect and high precision;but in the case of small sample,two methods has the same fitting effect.
出处
《水电能源科学》
北大核心
2012年第2期136-138,179,共4页
Water Resources and Power
基金
国家自然科学基金资助项目(51109180)
