人工智能在预测支持系统中的应用

On Developing Forecasting Support System in China

提出一种新型预测支持系统。预测分析方法采用基于径向基函数的神经网络方法 ,所有预测支持系统的数据放在数据仓库。针对不同预测支持子系统 ,通过在线分析处理 ( OLAP)及数据开采 ,从数据仓库中提取综合数据和信息 ,作为预测输入样本。该方法用于酒店物资部预测支持子系统 ,为部门主管提供预测信息 ,使其制定出合理的进货计划 ,保证了物品不短缺的同时也不积压 ,使宾馆管理系统高效。

For some cases, such as a hotel, it is difficult to establish the mathematical model for the forecasting support system needed. Fig.1 shows the schematic diagram of the forecasting support system we design and implement; its FSS (forecasting support system) program is RBF (radial basis function) neural network. In subsection 1.2 we discuss in much detail our RBF neural network. We employ RBF neural network for training with the stored data of FSS. In section 2, we take a hotel as example. Fig.4 shows the results of using RBF network to train with stored data; when the number of neurons is equal or greater than 10, the mean square error is small enough to satisfy hotel needs. Fig.5 shows that simulation results of our FSS are in good agreement with actual monthly consumptions of 40W incandescent lamps.