摘要
燃烧稳定性判定问题是锅炉燃烧状态自动监测中亟待解决的关键问题之一。为了实现自动化监测,并对燃烧的稳定程度进行量化判定,基于数字图像处理技术,从炉腔火焰图像中提取燃烧参数,建立燃烧参数数据库。基于多属性判定方法,生成区间数据样本决策库。在模糊推理中,为获取隶属度函数参数和模糊推理规则,提出了基于粗糙集简化样本决策库。按照决策属性离散化决策库中的条件属性,实现了属性简约和属性值简约,增加了网络训练样本参数的可靠性。结合模糊网络的逻辑推理性和神经网络的学习性、并行计算等优点,建立了用于燃烧诊断的T-S模糊神经网络模型。选择合适的模糊分割数,定义"五四模型",建立基于"五四模型"的火焰燃烧稳定性判定模型,并进行仿真试验。对比训练前后的仿真图参数表明,该模型是可行的,并具有较好的试验效果。
Stability determination of combustion is one of the key issues to be solved in the automatic monitoring of boiler combustion state. For the purpose of automatic monitoring and quantizing the combustion stability degree,according to the digital image processing technology,the combustion parameters are extracted from flame images of furnace cavity and the database of combustion parameters is built. Based on multi-attribute determination method,the interval number sample decision library is generated. In fuzzy inference,for obtaining the membership function parameters and fuzzy rules,the simplified sample decision database is proposed based on rough set. According to the condition attributes of discretized decision database,the simplicity of attributes and their values are realized,and the reliability of sample parameters of network training is increased. Combining the logical inference of fuzzy network and the advantages of neural network,such as good learning ability,parallel computing,etc.,a T-S fuzzy neural network model is built to diagnose combustion. Appropriate number of fuzzy divisions is selected; and ＂5-4Model＂is defined; and the determination model of flame combustion stability is built based on ＂5-4 Model＂; and simulation experiment is carried out. Contrasting the parameters of simulation diagrams before and after training,the results show that the model is feasible.
出处
《自动化仪表》
CAS
2017年第7期29-33,共5页
Process Automation Instrumentation
关键词
燃烧稳定性
火焰图像处理
样本决策库
隶属度函数
粗糙集
离散化
T-S模糊神经网络
分割数
五四模型
Combustion stability
Flame image processing
Sample decision database
Membership function
Rough set
Discretization
T-S fuzzy neural network
Number of divisions
5-4 model
