基于神经网络、模糊理论的变压器油中气体诊断专家系统

随着大容量变压器的应用日益广泛,以在线监测及故障诊断技术为基础的状态维修体系愈发显得重要,国内外已有诸如神经网络模式识别、基于不确定推理的模糊化处理等智能诊断技术,在各自擅长的领域内应用并取得了一定的成果,但单一的方法均存在不同程度的弊端。该文以人工智能理论中的模糊理论和神经网络理论为基础,介绍了面向对象技术的变压器油中气体诊断黑板型专家系统,该系统体现了专家系统启发性、透明性、高性能和灵活性的特点,不仅能够利用模糊理论来解决变压器故障诊断时出现的复杂性、经验性和模糊性等问题,还能利用神经网络的良好模式分类能力和自学习功能来不断地提高整个系统的故障诊断精度,从而有效地克服了以往采用单一方法进行诊断时所固有的诊断出错或诊断结果不全面的缺点,提高了电力变压器绝缘诊断的准确度,体现了该专家系统的优越性。

Transformer DGA Diagnosis Expert System Based on Neutral Network and Fuzzy Theory

As large capacity transformers are widely used,state maintainance based on inspectation on line and fault diagnosis becomes important.Neutral network and fuzzy theory have done their best in this area.However,single method has respective problem.This paper introduces a new kind of transformer DGA diagnosis blackboard expert system on the base of the fuzzy theory and the neutral theory.This System has the character of illumination,limpidity,high quality and agility.This system can not only use the fuzzy theory to solve the problem of the complexity,experiences and fuzziness when transformers have something wrong but also utilize the good patter sorting and self-learning abilities neutral network to improve the diagnosis precision of the whole diagnosis system.This system effectively overcomes the problem of wrong diagnosis or incomplete results based on single methods.So it can improve precision of transformer insulation diagnosis.