零维内弹道分析计算简便、高效,是一种常用的内弹道性能分析方法,但其常采用将喉径变化视为线性烧蚀的方法对喉部面积进行简化处理,而大量的试验表明喉径烧蚀并非线性,喉衬烧蚀主要发生在发动机工作的中后期。为此,文中提出了一种基于遗传算法和LSTM神经网络的内弹道性能预示方法,首先利用遗传算法和零维内弹道计算方程,从试验所得的压强数据中对喉径变化曲线进行辨识;然后利用LSTM神经网络建立喉径变化预示模型,实现对发动机喉径变化曲线的预示,并以此为基础对工作段内弹道性能进行预示;最后以实际案例对该方法进行验证,预示曲线与实际曲线的均方误差为0.118 MPa,相对误差为1.35%,结果表明该方法可以较为准确的预示固体火箭发动机的压强变化曲线。

Zero-dimensional internal ballistic analysis is a commonly used internal ballistic performance analysis method with simple calculation and high efficiency, and has a good performance in the study of internal ballistic performance. However, it often adopts the method of treating throat diameter change as linear ablation to simplify the throat area. A large number of analyses and tests show that throat diameter ablation is not linear, and throat liner ablation mainly occurs in the middle and late stages of engine operation. Therefore, a prediction method of internal trajectory performance based on genetic algorithm and LSTM neural network is proposed in this paper. Firstly, genetic algorithm and zero-dimensional internal trajectory calculation equation are used to identify the throat diameter change curve from the pressure data obtained from the test. Then LSTM neural network is used to establish a throat diameter change prediction model to predict the throat diameter change curve of the engine. On this basis, the ballistic performance in the working section is predicted. The method is verified by a real case.The mean square error between the predicted curve and the actual curve is 0.118 MPa, and the relative error is 1.35%. The predicted results verify that the method can accurately predict the pressure change curve of solid rocket motor.