基于改进主观贝叶斯方法识别电熔镁炉异常工况

电熔镁炉熔炼过程信息包含大量的不确定性，基于大数据分析的方法难以应用.为准确识别电熔镁炉熔炼过程的异常工况，提出一种基于改进的主观贝叶斯在线规则推理方法.传统的主观贝叶斯方法参数赋值范围过大，针对这一问题，使用映射函数将参数赋值范围缩小到有限区间，以提高方法的实用性.在规则推理时，使用模糊隶属度函数对观察和证据进行模糊匹配，以提高工况识别的鲁棒性和准确率.仿真分析表明该方法可以有效描述规则中的不确定性信息，准确识别电熔镁炉熔炼过程的异常工况.

Abnormal Condition Recognition Based on Improved Subjective Bayesian Method for Fused Magnesium Furnace

It is difficult for big data analysis to be applied to the smelting process of fused magnesium furnace because of a lot of uncertain information of the process. In order to identify abnormal conditions accurately， an online rule reasoning method based on improved subjective Bayesian is proposed. In view of the problem that the parameter value range of the traditional subjective Bayesian method is too wide， the mapping function is used to limit the value range to a finite interval， which improves the practicability of the method. In order to improve the robustness and accuracy of condition recognition， the fuzzy membership function is utilized to match the observation and evidence in the reasoning. Simulation results show that the method can effectively describe the uncertain information in the rules and accurately identify the abnormal conditions in the smelting process of fused magnesium furnace.