基于改进RBF的变压器油色谱在线监测装置配置策略研究

STRATEGY OF ONLINE MONITOR DEVICE ALLOCATION FOR TRANSFORMER OIL CHROMATOGRAM BASED ON IMPROVED RBF NEURAL NETWORK

变压器油色谱在线监测是变压器故障诊断的重要举措,也是电网变电检修的重要手段。考虑到110千伏(66千伏)电压等级油浸式变压器在线监测装置的选配对其诊断效果具有较大影响,并且其影响因素较多且复杂,提出一种基于K-means聚类和遗传算法相结合的RBF神经网络在线监测装置配置策略。从变压器内部本体特征和外部所处环境两个方面对装置选配影响因子进行筛选和分析得到最终的影响因子;对RBF模型的适用性进行分析和阐述;建立以影响因子为输入向量,以监测装置配置评分作为输出向量的基于改进RBF在线监测装置配置模型。实验结果表明,该模型能够对在线检测装置进行准确评分,提高了选配准确度,证明了该模型的有效性。与神经网络(BP)模型相比,该模型加快了网络收敛速度,能够更加有效地解决在线监测装置配置问题,为电网提供切实可行的方案。

Oil chromatography online monitor technology has become an important measure for transformer fault diagnosis,which is also a key means of transformer substation maintenance.The selection of online monitor device for 110 kV(66 kV)oil-immersed transformer has great influence on its diagnosis effect,and its influencing factors are many and complex.Therefore,this paper proposes an online monitor device allocation strategy based on the RBF neural network which combines both the K-means and genetic algorithm.The strategy filtered and analyzed the influencing factors of device selection from two aspects of internal ontology characteristics and external environment of transformer to get the final influencing factors.We analyzed and elaborated the suitability of RNF model.An improved RBF-based online monitor device allocation model was established,in which the impact factor was the input vector and the configuration score of the monitoring device was the output vector.The experimental results show that the model can accurately score the online monitor device allocation,improve the allocation accuracy,and prove the effectiveness of the model.Comparing it with the BP model,our model accelerates the convergence speed of the network and effectively solves the online monitor device location,which provides a feasible solution for power grid.