摘要
校准间隔优化研究是舰艇作战系统动态对准理论的重要组成部分;针对目前作战系统校准间隔确定缺乏理论支撑、历史对准误差数据使用率较低的情况,根据历史对准误差数据的特点,提出了利用EEMD-LSSVM-B的组合预测方法对对准误差数据进行预测,从而确定作战系统的校准间隔的优化方法;首先利用EEMD将对准误差数据分解成若干不同频率的分量,分解后的各分量通过LSSVM进行预测,LSSVM的相关参数由QDE算法优化获得,各预测分量通过BP神经网络进行非线性重构,得到最终的预测结果,最后根据预测值对校准间隔进行优化调整;仿真实例表明,组合预测方法比单独的预测方法具有更高的精度,可以应用于作战系统校准间隔的优化过程。
Calibration interval optimization is an important component of the dynamic alignment theory of combat system. For the lack of theoretical support of calibration interval determination of combat system and the low usage of historical alignment error data, a calibration interval optimization method utilizing EEMD-- LSSVM-- BP combination forecasting is proposed according to the characteristic of historical a- lignment error data in this paper. Firstly the alignment error data is decomposed into several different frequency components by EEMD, which is predicted by LSSVM, parameters of LSSVM are obtained by QDE optimization algorithm, and each prediction component are recon- structed nonlinearly by the BP neural network to give the final prediction, and the calibration interval is adjusted and optimized based on the predicted value at last. The simulation shows that the combination forecasting method has higher accuracy than the individual prediction method, and can be applied to the calibration interval optimization process.
出处
《计算机测量与控制》
2016年第12期96-99,共4页
Computer Measurement &Control
关键词
作战系统
校准间隔
集合经验模态分解
最小二乘支持向量机
量子差分进化
BP神经网络
combat system
calibration interval
ensemble empirical mode decomposition
least square support vector machine
quan- tum differential evolution
BP neural network
