摘要
提出一种简单而有效的从数据中提取模糊 IF- THEN规则的混合方法 :首先通过启发式方法确定模糊 IF- THEN规则结论部分非模糊单值 (即实数 )的初始值 ,然后通过梯度下降学习方法对模糊模型前提参数和结论参数同时进行精调。这种方法的优点是不仅模型精度较高 ,而且收敛速度快。由于提出的方法对模型的性能有明显的影响 ,因此 ,对如何系统地调整输入—输出数据对权值的方法也进行了研究。
It proposes a simple but powerful hybrid method for automatically generating fuzzy if-then rules from numerical data: firstly, the initial values of fuzzy if-then rules with nonfuzzy singletons (i.e., real numbers) in the consequent parts are generated by the heuristic method, then the gradient descent learning algorithm is used to precisely adjust premise parameter and consequent parameter simultaneously. The advantages of proposed method are not only the better approximation accuracy, but also the faster convergence speed. The proposed method has definite effects on the model's performance; therefore, the way to systematically adjust the weight of the input-output data pair is also investigated. Finally, the simulation result from an example is presented to demonstrate the superiority of the proposed model to the conventional methodologies.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2004年第3期382-384,417,共4页
Chinese Journal of Scientific Instrument
基金
国家杰出青年基金 ( 6992 5 3 0 8)
黑龙江省自然科学基金
哈尔滨工业大学科学研究基金资助项目
