摘要
本文给出了一种新的组合优化算法——模拟退火优化算法。对算法的机理、构造准则、参数选取及收敛性作了较深入的理论研究,给出了改进的退火算法;并对模拟电子设备的故障诊断问题作了讨论,给出了基于退火算法的电子线路故障诊断方法。
A new combinatorial optimization technique, called simulated annealing optimization
algorithm, has already proved to be a very powerful method that can be used in many areas
of research [4].
The authors three main contributions are stated and then detailed in the next three
paragraphs.
The authors extend the numerical annealing algorithm framework presented by
Kirkpatrick et al.[5] to the case of continuous multidimensional parameter space. In con-
trast, Kirkpatrick requires that the variables be limited to discrete space. We design an effi-
cient parameter inverse simulated annealing algorithm SA. This paper presents some im-
portant theoretical results on continuous annealing algorithm design: (a) the selection of
maximum and minimum temperature scales, (b) the sampling method of randomly gener-
ating new states, (c) the strategy of annealing schedule and especially(d) the mathematical
proof that SA converges to global optimum solution in probability.
The authors apply the SA optimization algorithm to fault diagnosis of analog circuits
with tolerance. We formulate a combinatorial optimal objective function. This function not
only conveys the faulty information thoroughly, but also takes into consideration the
combinatorial tolerance effects of non-faulty elements. Utilizing the SA algorithm as a
powerful tool in solving the formulated optimal problem, we can identify the faulty ele-
ments and evaluate tolerance effects of non-faulty elements simultaneously.
A software package for automatic diagnosis has been developed for the electronic
equipment of remotely piloted vehicles. This software package has been applied to fault iso-
lation of the electronic equipment of a remotely piloted vehicle developed by NPU Institute
of Pilotless Mini-Aircraft. The diagnosis results, obtained with the software package, both
in ground simulation and in flight testing, are rather satisfactory.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
1992年第2期188-196,共9页
Journal of Northwestern Polytechnical University
基金
航空科学研究基金资助课题
关键词
故障诊断
电子设备
飞机
退火算法
combinatorial optimization
simulated annealing
fault diagnosis
analog circuit
