多元线性回归和神经网络在水库调度中的应用比较研究

Comparative Study on Application of Multi-Variable Linear Regression and Neural Network in Reservoir Operation

采用多元线性回归和改进人工神经网络算法分别建立水库多目标调度函数 ,并进行实例计算验证。通过比较分析 ,表明人工神经网络的非线性映射能力能够更好地反映水库调度中多个自变量和因变量之间的复杂关系 ,具有较高的模拟精度和较好的可行性 ,且应用简便。

With applications of multi-variable linear regression and improved artificial neural network algorithm, multi-objective reservoir operation functions are established respectively and verified by case study. Through comparison and analyses, it is shown that the artificial neural network algorithm has a better non-linear reflection ability, and can describe the complex relationship between the independent variable and dependent variable with higher precision and better feasibility, and is easier to be used.