摘要
火炮初速决定着在复杂战场环境中能否准确打击敌人,而准确预测出火炮初速关系到在不经试射的情况下能否成功命中目标。预测火炮初速往往采用某种单一模型,虽然建模简单但是只能提取出火炮初速中的某一特征,从而导致预测精度并不理想。针对这种情况,选取了某型火炮3组不同的初速数据进行分析,提出利用ARIMA时间序列模型、GM(1,1)灰色模型及BP神经网络模型进行预测,既能提取出火炮初速中的线性成分又能提取出非线性成分,同时为了最大限度发挥出单一模型的预测优势,利用3个单一模型建立了组合模型,并利用实测数据对各个模型预测精度进行了检验。结果表明,组合模型能更好地发挥出所有模型的预测优势,预测精度更高,更适合作为火炮初速预测的有效模型。
The gun muzzle velocity determines the ability to accurately strike the enemy in a complex battlefield environment,and the accurate prediction of the gun muzzle velocity is related to the success of hitting the target without test firing.A single model is often used to predict gun muzzle velocity.Although the modeling is simple,only a certain feature of gun muzzle velocity can be extracted,so the prediction accuracy is not ideal.To address this situation,three different groups of muzzle velocity data of a certain type of gun were selected for analysis,and it was proposed to use ARIMA time series mo-del,GM(1,1)gray model and BP neural network model for prediction,which can extract both linear and nonlinear components of the muzzle velocity of the gun.And at the same time,in order to maximize the advantages of a single model for prediction,a combination model was established based on three single models,with the prediction accuracy of each model tested by using the measured data.The results show that the combined model can better exploit the prediction advantages of all the mo-dels,with higher prediction accuracy,and is more suitable as an effective model for gun muzzle velocity prediction.
作者
田珂
陈铎
TIAN Ke;CHEN Duo(Unit 63850 of PLA,Baicheng 137001,Jilin,China)
出处
《火炮发射与控制学报》
北大核心
2021年第1期30-35,共6页
Journal of Gun Launch & Control