基于BA-RBFNN控制图模式识别的气化过程参数失稳监控模型  

Monitoring model of gasification process parameter instability based on BA-RBFNN control chart pattern recognition

作　　者：张自川 张进春[1] ZHANG Zichuan;ZHANG Jinchun(School of Energy Science and Engineering,Henan Polytechnic University,Jiaozuo 454000,Henan,China)

机构地区：[1]河南理工大学能源科学与工程学院,河南焦作454000

出　　处：《河南理工大学学报（自然科学版）》2021年第4期38-47,共10页Journal of Henan Polytechnic University（Natural Science）

基　　金：国家自然科学基金资助项目(51774113)。

摘　　要：为识别固体燃料气化过程参数出现的异常模式(失稳),构建一种基于蜜蜂算法-径向基函数神经网络(BA-RBFNN)控制图模式识别的气化过程参数失稳监控模型,对气化过程参数进行监控。该监控模型主要包括特征描述、特征选择、分类器和训练方法4个模块。选择形状特征和统计特征对气化过程参数进行描述,运用关联规则算法(AR)选择最佳特征集合,选择径向基函数神经网络(RBFNN)作为分类器,采用蜜蜂算法(BA)作为模型的训练方法。为检测模型性能,用模拟数据和气化炉现场数据分别对模型进行测试,并与传统方法对比。结果表明,该模型对气化过程参数出现的异常模式具有更好的识别监控效果。To identify variation patterns in the process of gasification of solid,a monitering model based on bees algorithm-radial basis function neural network(RBFNN)was proposed to conduct pattern recognition of the parameters of gasification process.The model consisted of four modules:feature description,feature extraction,classifier and training method.Shape feature and statistical feature were selected to describe gasification process parameters.Association rule(AR)algorithm was nsed to select the best feature set.Radial basis function neural network(RBFNN)was selected as classifier.Bees algorithm(BA)was used as the training method of the model.In order to test the performance of the model.The simulation data and the gasifier field data were used to test the model respectively,and the pnodel was compared with the traditional method.The results showed that the proposed model had a better recognition effect on the abnormal patterns in the gasifier parameters.

关 键 词：气化过程参数 模式识别 径向基函数神经网络 蜜蜂算法 关联规则算法 

分 类 号：F253.3[经济管理—国民经济] X937[环境科学与工程—安全科学]