摘要
在预定的温度下,飞机失事后机身摩擦受热时间与燃烧的关系很复杂,受到飞行高度、天气状况等综合因素的影响,是一个多变耦合数学问题。针对飞机失事后机身摩擦受热时间与燃烧的数学建模,需要设定大量的约束条件,缺少智能化过程,导致建模结果不准确,提出了一种新的飞机失事后机身摩擦受热时间与燃烧的数学关系模型。依据机身摩擦受热模型,采用Themecmastor-Z型热加工模拟试验机对机身进行不同程度的摩擦受热试验,依次记录各个时间段以及各种温度下相关机身摩擦受热的燃烧数值,将该数值输入神经网络系统中,建立前馈式算法的人工神经网络,并进行优化训练,构造出飞机失事后机身摩擦受热时间与燃烧的数学关系模型。实验结果说明,与传统模型对比,改进的神经网络模型具有一定高效性,将误差缩小到3%,大大提高了模型自身的拟合度,精确地描绘出飞机失事后机身摩擦受热时间与燃烧的数学对应关系。
A new mathematical relationship model of fuselage friction heating time and burning after the crash was proposed in the paper. According to the fuselage friction heating model, different degrees of friction heating tests of the fuselage were performed by using the Themecmastor - Z thermal simulation testing machine. And the related burning values of the fuselage friction heating in each time period and all kinds of temperatures were successively re- corded. Then, the values were input to the neural network system to establish the artificial neural network based on feedtbrward algorithm and make optimizing training, so as to construct mathematical relationship model of the fuselage friction heating time and the burning after the crash. Experimental results show that, compared with traditional mod- el, the improve neural network model has a certain efficiency, and reduce the error by 3% , which greatly improves the degree of fitting of the model itself and accurately describes the mathematical corresponding relation of the fuse- lage friction heating time and burning after the crash.
出处
《计算机仿真》
CSCD
北大核心
2015年第6期84-87,共4页
Computer Simulation
基金
国家自然科学基金(11301291)基于sieve bootstrap方法的长记忆过程变点研究与应用
关键词
机身受热摩擦
摩擦受热时间
燃烧数值
Fuselage friction heating
Friction heating time
burning value
