| 摘要: |
| 对于具备节流能力的液体运载火箭弹道优化问题，不同于传统动力弹道解耦设计方法，为开展考虑节流后比冲变化影响的实时弹道节流优化设计，构建耦合发动机流量特性的动力弹道一体化精细模型，实现运载能力的精准评估。为解决精细化模型优化时间效率问题，将径向基神经网络近似模型应用到实时动力弹道耦合仿真中，单轮优化时间降低81.4%，且精度误差小于1%。此外，基于径向基神经网络模型的近似优化算法在显著缩短寻优时间的同时，还具备通用移植性，在实时在线优化等工程领域应用前景广阔。 |
| 关键词: 液体运载火箭 动力弹道耦合设计 近似优化 径向基函数 神经网络 |
| DOI： |
|
| 基金项目:火箭创新基金（ZH2022003） |
|
| Research on Coupled Optimization Design of Liquid Launch Vehicle Propulsion and Trajectory Based on Radial Basis Function Neural Network |
| HAO Wenzhi,ZHANG Zhiguo,ZHU Hao,HE Wei |
| (1.School of Astronautics, Beihang University, Beijing 100191, China;2.Beijing Institute of Astronautical Systems Engineering, Beijing 100076, China) |
| Abstract: |
| For the trajectory optimization problem of liquid launch vehicles with throttling capability, unlike traditional propulsion and trajectory decoupling design methods, in order to carry out real-time trajectory throttling optimization design considering the influence of specific impulse changes after throttling, a coupled engine flow characteristic propulsion and trajectory integrated fine model is constructed to achieve accurate evaluation of launch capability.To address the issue of time efficiency in optimizing refined models, the radial basis function (RBF) neural network approximation model was applied to real-time propulsion and trajectory coupling simulation. This approach led to a significant reduction of 81.4% in single round optimization time, with an accuracy error of less than 1%. Moreover, the approximate optimization algorithm based on the RBF neural network model significantly shortened the optimization time and exhibited universal portability, offering broad prospects for engineering applications, such as real-time online optimization. |
| Key words: Liquid launch vehicles Coupled propulsion and trajectory design Approximate optimization Radial basis function Neural network |