自适应神经模糊推理网络在微波效应预测中的应用

Application of adaptive neuro-fuzzy inference system to forecast of microwave effect

针对微波效应的预测问题,采用自适应模糊神经网络的方法,以微波参数和作用阈值为参变量,初步建立基于自适应神经模糊推理网络的微波效应预测模型,并引入置信水平、置信区间等概率统计概念对模型的预测能力进行评价。通过学习样本和检验样本与模糊预测输出的对比结果表明:自适应神经模糊推理网络对样本需求少,预测能力强,泛化误差小,具有一定的应用推广价值。

In order to solve the forecast problem of the microwave effect, the adaptive neuro-fuzzy inference system （ANFIS） is applied and the microwave parameters and its thresholds are adopted as variables. As a result, the ANFIS model based on the microwave effect data is developed. The confidence level and confidence interval are proposed to appraise the forecasting ability of the ANFIS model. Comparing the output of the ANFIS with the training and testing data, good forcasting ability, small error and low data requirement are found with the ANFIS.