基于神经网络的惯性平台温度场仿真模型修正方法

Simulation model correction method of inertial platform temperature field based on neural network

针对目前平台系统仿真方法计算效率低且精度不高的问题,提出了一种基于神经网络算法且考虑真实壁面条件的惯性平台系统仿真模型修正方法。从惯性平台三维仿真模型入手,使用神经网络算法修正后的对流换热系数作为惯性平台与外界的换热条件,建立系统与无限大空间进行热交换的仿真模型,利用该仿真模型得到了更高精度的系统内部温度分布。仿真与试验结果表明,修正后的仿真模型中网格数量减小24.61%,仿真精度提高了15.95%。所得到的对流换热系数修正系数可用于具有相似外壁面惯性平台的温度场仿真分析。

Aiming at the problem of low efficiency and low accuracy of the current platform system simulation method, a simulation model correction method of inertial platform system is proposed based on the neural network algorithm and taken into account the real wall conditions. Starting from the inertial platform＇s 3 D model, the modified heat transfer coefficient by the neural network is used as the heat exchange condition between the system and the external environment, and a simulation model for the heat exchange between the system and the infinitely large space is established. Higher-precision temperature distribution within the system is obtained by using this model. Simulation and test results show that the number of elements in the new simulation model with modified coefficients is decreased by 24.61%, and the simulation accuracy is increased by 15.95%. In addition, the modified convective heat transfer coefficient can be used in the temperature field simulation analysis of inertial platforms with similar external wall.