At present, the optimization design method of electromagnetic coil launcher mainly includes of manual trial and error method and intelligent algorithm. The former has low optimization efficiency and poor effect, while the latter can obtain optimization results, but the calculation time is longer. In order to improve the calculation efficiency and obtain the global optimal solution of the coil launcher, the optimization process of the coil launcher must be improved. In this article, a genetic algorithm based on a predictive model is adopted. By using the output of the predictive model as the optimization target and constraint target of the optimization algorithm, it avoids calling the full model of the coil launcher during the optimization process and effectively reduces the calculation time. The orthogonal test method and the current filament method are used to establish 49 sets of orthogonal sample data of the single-stage coil launcher. The optimal armature speed of the coil launcher obtained in the orthogonal test is 59.81 m/s; the support vector regression (SVR) prediction model is obtained by training on the orthogonal test results, and the prediction error of the test data is 0.39%; the genetic algorithm based on the prediction model is used to optimize the single-stage coil launcher. The calculation result shows that, without constraints, the peak speed of the armature is 60.17 m/s. After the temperature constraint is added, the peak speed of the armature is 59.86 m/s.

中文翻译：

---

基于预测模型的单级线圈发射器优化计算


目前，电磁线圈发射器的优化设计方法主要包括人工试错法和智能算法。前者优化效率低、效果差，后者虽能得到优化结果，但计算时间较长。为了提高计算效率并获得线圈发射器的全局最优解，必须改进线圈发射器的优化过程。本文采用基于预测模型的遗传算法。通过将预测模型的输出作为优化算法的优化目标和约束目标，避免了优化过程中调用线圈发射器的完整模型，有效减少了计算时间。采用正交试验法和电流灯丝法建立了单级线圈发射装置的49组正交样本数据。正交试验得到的线圈发射器最佳电枢速度为59.81 m/s；对正交测试结果进行训练得到支持向量回归(SVR)预测模型，测试数据的预测误差为0.39%。采用基于预测模型的遗传算法对单级线圈发射器进行优化。计算结果表明，在没有约束的情况下，电枢的峰值速度为60.17 m/s。加入温度约束后，电枢的峰值速度为59.86 m/s。