摘要
针对传统的锅炉故障诊断方法中神经网络模型结构复杂、信号繁多、训练时间长等缺点,提出了一种基于FAHP和ANN结合的风险评估研究方法;采用FAHP分析锅炉的安全层次结构,通过对影响锅炉安全状态的若干因素之间隶属度的判别,构造出反映人类专家经验与客观事实一致性的模糊矩阵,进而定量地计算出各因素的权重系数;选择权重系数较大的因素作为锅炉安全ANN模型的输入,从而得到锅炉的安全层级;经实际验证,此方法既保留了关键信息,又剔除了冗余信息的干扰,从而简化了ANN的结构,缩短了运算时间,在保持评估准确性的前提下,满足了锅炉安全评估快速性的要求。
Aiming at disadvantages of traditional boiler fault diagnosis method, which includes complicated ANN model structure, multiple signals and long--time training, this paper brings forward a risk assessment research method based on the combination of FAHP and ANN. The method adopts FAHP to analyze boiler＇ s security hierarchy structure, and by distinguishing membership degree of influential factors of boiler＇ s safety state, it constructs a fuzzy matrix which reflects the consistency between human expert＇ s experiences and objective facts. Prior to this, the method figures out each factor＇ s weight coefficient. By selecting factors of bigger weight coefficient as input of boiler safe ANN model, it gets security level of the boiler. Through actual test and verify, this method not only keeps key information, but also eliminate the disturbance of redundant information. Accordingly, its simplifying ANN＇ s structure and shorten of the calculation time so that satisfies the requirement of instant boiler＇ s security assessment under the premise of preserving assessment veracity.
出处
《计算机测量与控制》
CSCD
北大核心
2011年第2期392-394,共3页
Computer Measurement &Control
基金
南京工业大学学科基金
江苏省工业装备数字制造与控制技术重点实验室开放基金(BM2007201)资助
关键词
模糊层次分析法
人工神经网络
锅炉f风险评估
FAHP （fuzzy analytic hierarchy process）
ANN （artificial neutral network）
boiler~ risk assessment
