摘要
针对以往变压器油中溶解气体含量预测中对所有数据样本等同处理,没有区分不同时间样本对建模作用不同而造成预测误差的问题,提出一种基于模糊支持向量机预测模型,将样本按照时间由近及远赋予不同的权重,并采用自适应遗传算法优化其参数,根据适应度值自动调整交叉概率和变异概率,保证参数的全局最优性,克服参数选择的盲目性。实例分析证明,该模型应用于变压器DGA含量的预测中,有效降低了预测误差,提高了预测精度。
At present,all data samples are treated equally in prediction model of dissolved gases in transformer oil.Different time samples make no difference on modeling inevitably and produces prediction errors.To solve this problem,a novel prediction model based on fuzzy support vector machines is proposed in this paper.Each input sample is assigned to different weights according to its sampling time.And parameters in FSVM model are optimized by adaptive genetic algorithm.Crossover probability and mutation probability are adjusted automatically according to the fitness values of the object function.It ensures the global optimization of parameters and overcomes drawbacks of blindness in selection.Simulation results show that the proposed forecasting model decreases prediction error and improves prediction precision greatly.
出处
《中国农村水利水电》
北大核心
2012年第7期167-170,共4页
China Rural Water and Hydropower
关键词
变压器
油中溶解气体
自适应遗传算法
模糊支持向量机
预测
transformer
dissolved gases in oil
adaptive genetic algorithm
fuzzy support vector machines
prediction
