摘要
通过对常规最小方差型目标函数局限性的分析 ,根据鲁棒统计学理论和目标函数在参数学习中的导向作用 ,对目标函数进行修正。在此基础之上 ,提出一种模糊逻辑系统的鲁棒学习算法。在噪声环境中 ,通过对该算法的仿真验证以及与常规算法性能的比较 ,表明该鲁棒学习算法在逼近精度和鲁棒性等方面优于传统方法 ,在实际工程中具有较高的应用价值。
According to robust statistics theory and the directional role of target function, the limitation of the LSE (Least Squared Error) function is analyzed. A new robust learning algorithm of fuzzy logic systems is presented. Simulation results and the performance comparison between the robust learning algorithmandthenormalalgorithm,showthattherobustlearningalgorithmhasbetterrobust capability, and has more practicality engineering applications.
出处
《控制与决策》
EI
CSCD
北大核心
2003年第1期69-72,共4页
Control and Decision