摘要
提出小生境均匀变异算子,用共享度来决定个体的变异概率,以竞争方式接受变异结果,克服了遗传算法的早熟问题.采用GA-BP混和训练法建立的阻力数值图谱具有较高的估算精度;利用遗传算法建立的船型参数优化方法,能够得到设计范围内的最优解,使设计的滑行艇阻力最小.
Niche uniform mutation operators are brought forward to get over genetic algorithms’ earliness. Individual mutation rate depends on its sharing value, and mutation result is accepted according to its value sharing and fitness. GA-BP hybrid training method is used to form resistance numerical atlas to acquire higher estimating precision. Genetic algorithms are used to optimize ship form parameters to get minimum resistance planing boat hull form.
出处
《船舶工程》
CSCD
北大核心
2005年第3期15-19,共5页
Ship Engineering
关键词
船舶
神经网络
遗传算法
滑行艇
ship
neural network
genetic algorithm
planing hull
