基于粒子群聚类的导弹直接侧向力模型辨识

为了充分挖掘数据的潜在规律,提出一种基于粒子群模糊C均值聚类的模型辨识方法,该方法首先对实验数据进行分类,得到最佳聚类数和聚类中心,然后对不同的类分别辨识.新数据可以根据到不同聚类中心的距离划分为不同的类,再根据该类的辨识模型得到输出结果.依据风洞实验数据,采用粒子群模糊聚类对直接侧向力模型进行了辨识,辨识结果表明,与直接辨识相比,该方法的辨识精度明显提高,其辨识结果可以为控制系统的设计提供依据.

Missile Lateral Jet Model Identification Based on Particle Swarm Clustering

To mine the rule of data sufficiently,a model identification method based on particle swarm clustering is provided.Experiment data is clustered with this method to gain best partition and clustering centers,and then different class is identified respectively.If the distance between new data and some clustering center is closer than others,and the new data is merged in corresponding class,through whose model the output of new data is obtained.The model’s inputs and output is from the wind channel experiment results.Through identifying missile lateral jet model using this method,result shows that its precision is improved apparently than traditional identification model.The Identification result based on particle swarm clustering can supply support for the design of control system.