主元分析优化量子神经网络的变压器故障诊断

Transformer fault diagnosis based on principal component analysis optimized quantum neural network

针对变压器故障检测速率较慢的问题,通过对基于量子神经网络的变压器故障诊断方法的分析,发现该方法有较高的精度,但是速率较慢,不能达到实时性、快速性的要求。因此提出基于主元分析优化量子神经网络的变压器故障诊断方法。利用主元分析进行故障数据降维,选取主成分累计贡献率高于85%的主元代替原有的7个故障气体含量数据,用降维后的数据作为网络的输入,应用量子优势消除数据相关性,最终对变压器的故障做出判断。利用变压器故障实验数据信息库中的故障数据分别对量子神经网络、主元分析优化量子神经网络进行仿真研究,结果表明在故障识别率不变的情况下,所提方法使得诊断速率得到大幅提升。

For transformer failure detection rate is slow,the transformer fault diagnosis method based on quantum neural network is analyzed. It is found that the method had high diagnosis accuracy,but the speed is slow,and can not meet the real time and rapid requirements. Therefore,a transformer fault diagnosis method based on principal component analysis and optimized quantum neural network is proposed. The principal element analysis is used to reduce the dimension of the fault data, and the principal element with a principal component cumulative contribution ratio higher than 85% is selected to replace the original seven fault gas content data. The reduced dimension data is used as the input of the network. The quantum advantage is utilized to eliminate the data correlation and the transformer′s failure is ultimately determined. The quantum neural network and principal component analysis and optimization quantum neural network were simulated by using the fault data in the transformer fault experimental data information base. The results show that the method′s diagnostic rate is greatly improved with the same fault recognition rate.