改进贝叶斯算法在配网故障诊断中的研究

Research on Improved Bayesian Algorithm in Fault Diagnosis of Distribution Network

针对智能电网故障诊断中所存在的大量不确定信息的问题,本文对贝叶斯网络算法在电网故障诊断中的应用进行了相关研究.在原有Noisy-or、Noisy-and两类节点模型和线路、母线和变压器三类故障诊断模型的基础上,对传统贝叶斯算法的不足之处进行改进,提出了一种改进贝叶斯算法.最后以某地区实际配网为研究实例,结果表明,与传统方法相比,改进后的算法在故障诊断中具有更高的灵敏度,降低错误概率,可以构造最小错误概率的贝叶斯分类器等优点,在实际电力系统故障诊断中半角将有很好的应用前景.此外,本文通过图形用户界面(GUI),显示可疑故障元件、后验概率、误动拒动的保护断路器等信息,方便现场工作人员及时做出决策,进一步处理故障.

For the large amount of uncertain information in the fault diagnosis of smart grid, the Bayesian network fault diagnosis methods are studied. Based on the original Noisy-or, Noisy-and node models and fault diagnosis models of power lines, buses and transformers have been established. The improved Bayesian algo- rithm is proposed to improve the traditional Bayesian algorithm. It can be concluded that the improved algo- rithm has the advantages of higher sensitivity and less error probability in the fault diagnosis, and can con- struct the Bayesian classification of the minimum error probability by comparing with the traditional algo- rithm. And it will have a good application prospect in the actual power system fault diagnosis.In addition, the results can be displayed by graphical user interface（GUI） to facilitate the operator to make rapid decisions, which would provide practical help for power system operators.