基于BP神经网络实现的B样条插值在系统动力学模型的应用

在系统动力学建模的过程中,系统变量之间的关系难以确定,传统的建模方法存在着很大的主观性,利用BP神经网络的方法则可以避免这个问题。而在建立BP神经网络过程中需要大量的学习样本,然而通常采集到的数据往往是不足的。用3次B样条函数对历史数据进行插值的方法构建系统中状态变量的学习样本,解决BP神经网络模型中学习样本不足的缺点,能更好的反应变量之间的非线性映射关系。

Application of B-spline Interpolation in System Dynamics'Model Based on BP Artificial Neural Networks

As the process of system dynamics modeling, the system is difficult to determine the relationship between variables, the traditional method of modeling is a considerable subjectivity, the using of BP neural network method is to avoid this problem. It needed plenty of historical data to establish artificial neural networks modeling, the historical data was usually not enough, so 3-order B-spline function was applied to construct learning samples by interpolating historical data. It can figure out shortcomings of inadequate samples in BP Neural Network model, do better respond to non- linear mapping relationship between variables.