由于服役环境变化剧烈和影响因素复杂,火箭炮状态变化具有非平稳、非线性的特点,为更准确对其进行预测,提出将模糊综合评判法和NARX神经网络相结合的方法,以某型火箭炮发火回路为例进行状态预测实验。结果表明,经过对影响因素进行模糊综合评分,NARX神经网络的规模减小且收敛速度提高,对火箭炮状态变化趋势的预测也更加精确。

The state change of the rocket launcher is non-stationary and non-linear. In order to predict the state of rocket launchers more accurately, this paper proposes a prediction method combining fuzzy comprehensive evaluation method and NARX neural network. Taking the firing circuit of a certain type rocket launcher as an example, the prediction experiment shows that after fuzzy comprehensive evaluation of the influencing factors, the NARX neural network converges faster during training and predicts the state change trend of the rocket launcher more accurately.