一种基于状态特征的航天发射故障诊断技术

文中采用模糊聚类分析的方法,利用数据库所积累的历史故障数据,有效地实现了实时发生的故障与历史故障的聚类,在此基础上找出发生故障的仪器等部器件。主要做法是：首先建立了模糊聚类分析模型G,及其μg（ui）,ui∈U;其次给出了计算相似系数sij的欧几里得公式,并由此建立了模糊相似矩阵R,利用改进的模糊聚类方法进行关系演算,则可得到不同的＂类＂,对此,设计了一种基于状态特征的算法FJLA;最后可从与要诊断的故障相似的类中找出可能发生故障部器件。经实验分析,应用上述理论可将故障定位在较小的范围,往往是一两件部器件,可有效地辅助人们科学决策。对某型火箭测试应用,其有效率可达65%,对缩短航天发射时间及提高作战应急反应能力具有重要的理论和实用价值。

A Technology in Fault Diagnosis of Spaceflight Launch Based on State Character

Adopt a method of fuzzy clustering analysis,using historical fault data in database,implement effectively clustering between Real-time fault and historical fault,on a base of this,find out apparatus whichis going wrong.The main method is： firstly,set up a model of fuzzy clustering analysis G～ and μg（ui）,ui∈U;secondly,Euclid formula for calculating parallel coefficient sij,set up R,a fuzzy parallel matrix by this.Process relation figure by using improved method of fuzzy clustering,gained different＂genus＂.Scheme out an algorithm
FJLA based on state character; finally, find out possible apparatus which is going wrong from parallel genus with pending diagnostic fault. By experiment,applying this theory can pitch fault in a small area, sometimes one or two pars, it can assist people＇s scientific decision availably. To a certain rocket test application, its efficient can reach to 65 %, it has important theoretic and applied value for shortening time of spaceflight launch, enhancing response ability when meeting an emergency in campaign.