【摘要】 以深度学习为基础,构建了一个基于深度神经网络的回归预测模型。根据样本数据,以损失函数最小为训练目标,通过反向传播算法以及随机梯度下降方法训练模型,采用训练结果对高超声速飞行器表面热环境进行评估。采用典型飞行任务中的782个样本数据进行算例分析,预测结果与实际数据的对比结果表明,壁面热流峰值和温度峰值的平均误差分别为3.03%和1.86%,表明该模型具有快速、高精度的特点。更多还原

【Abstract】 Based on deep learning, this paper constructs a regression prediction model of deep neural network. According to the sample data, with the minimum loss function as the training target, the model is trained through the back-propagation algorithm and the stochastic gradient descent method. The training results are used to evaluate the thermal environment on the surface of hypersonic vehicles. An example is analyzed using 782 sample data from a typical flight mission. Comparing the predicted results with the actual data, the average errors of peak wall heat flux and peak wall temperature are 3.03% and 1.86% respectively, indicating that the model is fast and accurate.更多还原