基于在BP嵌入动量因子的IDSS函数逼近器

 在领域工程中,经验模型都有一定的适用范围和一定的时效性,这主要反映在参数和因子的选取上.加入动量项的BP算法能克服原来BP算法的缺点,运用于IDSS的知识推理、导则方面,能使模型的系数随环境的变化而变化.上述思想在“单流域多水库联合调度系统”和“云南城市圈域经济辐射力模型的构建及实证研究”的推理机的设计实验阶段中得到体现.把该算法用MATLAB和C++做成函数逼近器来进行曲线拟合,实验数据和结果证明,该改进的BP算法能自动对“新安江模型”和辐射力模型的参数进行调整使之适合不同时期段的情况,这大大加强了该模型对环境的适应性,将进一步提高该软件产品的通用性.

IDSS functional approximation based on BParithmetic mixed with momentum factor

In field engineering,experienced model all have their own scope of application,and period of effectiveness.These are determined by the selection of parameter and factor.The BP arithmetic mixed with momentum factor overcomes the weakness of pure BP arithmetic,and it is by the using of the knowledge speculation or deduction.The above-mentioned idea was embodied in the software of "Single Watershed and Diversified Reservoir Control System".A functional approximation was made by using MATLAB and C++.It automatically adjust the parameter of "New Anjiang model" to fit the different situation of different reservoir.So the parameter of model will change with the transformation of the environment,and greatly improve the flexibility of the model and the popularity of the software product.