一种确定飞灰含碳量ANN模型输入参数的方法

A Method on the Determination of Input Parameters of Fly Ash Carbon Content ANN Model

为使用尽量少的输入参数使ANN(Artificial neural network)模型达到较高的预测精度,利用某电厂运行数据,通过不同参数组合的灵敏度分析,探讨了不同参数组合对基于人工神经网络的飞灰含碳量预测精度的影响。结果表明:通过灵敏度分析,能够确定既满足小数目输入参数,又满足较高预测精度的最终输入参数组合。用精简后的输入参数可以实现对飞灰含碳量的准确预测。

In order to make the ANN（Artificial neural network）model a higher prediction accuracy with less possible input parameters.With operation data of a power plant,through different parameter combinations of sensitivity analysis,the effect of the combination of different parameters on artificial neural network prediction accuracy for fly ash carbon content is discussed in this paper.The results show that：through the sensitivity analysis,the final input parameter combination could be determined with less input parameters and high prediction accuracy.The streamline input parameters may be sufficient to obtain accurate fly ash carbon content.