用模块化模糊神经网络模型进行参数学习专家系统的研究是根据不同的炉况类型及其相关变量将整个参数学习模型分解为几个结构相似的子模糊神经网络模型。其中，子神经网络模型能够模拟诊断推理过程，动态地进行特征数据的权重、模糊等级的界限值和规则可信度的调整。采用该方法对现场采集的数据进行实验，结果表明该方法具有较好的学习效果。 The research of parametric learning expert system based on modular fuzzy neural network model is based on different furnace types and their related variables to decompose the entire parametric learning model into several structurally similar sub-fuzzy neural network models. Among them, the sub-neural network model can simulate the process of diagnostic reasoning, the weight of the feature data dynamically, the threshold of the fuzzy level and the adjustment of the rule credibility. Using this method to test the data collected in the field, the result shows that this method has a good learning effect.