摘要
将基于改进遗传算法(IGA)和误差反向传播(BP)算法相结合构成的IGA-BP混合算法用于训练神经网络。该混合算法有效克服常规BP和传统GA算法独立训练神经网络的缺陷,并应用于电力变压器溶解气体分析的智能故障诊断。实验诊断结果表明,IGA-BP混合算法的收敛速度快于BP及GA算法,并且具有较高的诊断准确率。
The hybrid algorithm based on improved genetic algorithm（IGA）and error backpropagation（BP）algorithm composed is used for training neural network.The hybrid algorithm IGA-BP can effectively overcome the defects which conventional BP and traditional GA algorithm train neural network independently and is applied in the intelligent fault diagnosis for dissolved gas analysis of power transformer.The experimental diagnosis results show that the convergence rate of IGA-BP algorithm is faster than BP and GA algorithm and has high diagnosis accuracy.
出处
《电气传动自动化》
2010年第5期48-51,共4页
Electric Drive Automation
基金
江苏广播电视大学学术带头人基金资助项目
