现代计算机(普及版)
現代計算機(普及版)
현대계산궤(보급판)
MODERN COMPUTER
2013年
9期
3-6,11
,共5页
支撑向量机%遗传算法%机器学习%个体适应度%光伏阵列
支撐嚮量機%遺傳算法%機器學習%箇體適應度%光伏陣列
지탱향량궤%유전산법%궤기학습%개체괄응도%광복진렬
Support Vector Algorithm%Genetic Algorithm%Machine Learning%Individual Fitness%Photovoltaic Array
提出一种结合支撑向量机的混合遗传算法S-GA,该算法基于支撑向量机训练获得较准确的个体适应度判别器,利用适应度判别器降低算法运算量,保持进化种群的多样性并控制进化向更优方向进行。 S-GA和标准GA相比,具有收敛速度快,搜索最优解效率高的优势。利用GA和S-GA算法对非规则墙体上光伏阵列设计问题进行仿真实验,发现该算法收敛速度相比GA提高3倍,搜索得到的最优解对应的适应度增大约10%。
提齣一種結閤支撐嚮量機的混閤遺傳算法S-GA,該算法基于支撐嚮量機訓練穫得較準確的箇體適應度判彆器,利用適應度判彆器降低算法運算量,保持進化種群的多樣性併控製進化嚮更優方嚮進行。 S-GA和標準GA相比,具有收斂速度快,搜索最優解效率高的優勢。利用GA和S-GA算法對非規則牆體上光伏陣列設計問題進行倣真實驗,髮現該算法收斂速度相比GA提高3倍,搜索得到的最優解對應的適應度增大約10%。
제출일충결합지탱향량궤적혼합유전산법S-GA,해산법기우지탱향량궤훈련획득교준학적개체괄응도판별기,이용괄응도판별기강저산법운산량,보지진화충군적다양성병공제진화향경우방향진행。 S-GA화표준GA상비,구유수렴속도쾌,수색최우해효솔고적우세。이용GA화S-GA산법대비규칙장체상광복진렬설계문제진행방진실험,발현해산법수렴속도상비GA제고3배,수색득도적최우해대응적괄응도증대약10%。
Presents a S-GA (support vector machine - genetic algorithm), a hybrid genetic algorithm, which combined genetic algorithm with SVM algorithm. By SVM training, the S-GA generates an arbiter of individual fitness which used to reduce the amount of algorithm operation and con-trol the evolution to better direction as well as maintain the diversity of evolutionary population. Compared with the standard GA, S-GA converges faster and searches for the optimal solution more efficient. Uses the GA and S-GA on irregular wall PV array design by MatLab simulation respectively, finds that the proposed algorithm convergence speed is about 3 times of GA, and the fitness of the optimal solution increases about 10%.