基于AHP-BP神经网络法的石油化工企业安全性预警研究

Research on Safety Warning of Petrochemical Enterprises Based on AHP-BP Neural Network Method

结合石油化工企业加氢裂化装置运行过程中的危险因素,建立了安全性预警指标体系,初选出4个一级指标和33个二级指标,应用层次分析法计算各级指标权重,通过区间估计筛选二级指标,得到31个二级指标,对其使用隶属度函数进行标准化处理,建立了基于层次分析和区间估计的危险度评价方法。在此基础上依据BP神经网络建立石油化工企业安全性预警模型,通过多目标加权函数计算预警警度,并划分了预警警度与警度界限。以某石油化工企业为例,应用MATLAB软件对预警过程进行计算,得到了无警的预警结果,与实际结果相对比,模型的预测性能较优异且预测精度较高,并依据其指标危险性提出了针对危险指标的安全对策。

The comprehensive security early warning index system was established combined with petrochemical enterprise risk factors in the process of hydrocracking unit operation and the four level indicators and 33 secondary indexes were selected.The analytic hierarchy process were used to calculate index weight at all levels,the secondary indexes were screened by interval estimation to obtain 31 secondary indexes,which were standardized by membership functions,and the risk evaluation method based on hierarchy analysis and interval estimation was established.The integrated petrochemical enterprise safety early warning model was established based on BP neural network,and the warning degree was calculated by multi-objective weighted function,and the early warning degree and boundaries were divided.Taking one petrochemical enterprise as an example,the early warning process was calculated by MATLAB software,and the early warning result of light alarm was obtained.Compared with the actual results,the prediction performance of the model was better and the prediction accuracy was higher,and the safety countermeasures were put forward according to the index risk.