基于贝叶斯网络的电力变压器状态评估

为提高电力变压器状态评估的准确性,提出了一个基于贝叶斯网络的电力变压器状态评估模型。该法将变压器分为本体、套管、铁心3个部件,采用5级状态的评估方法,针对变压器预防性试验数据,先建立变压器健康状态量化的分层模型,通过该模型评估变压器的历史、当前、未来状态,然后利用模糊隶属度函数确定分层模型中变压器各个参数的阈值和分值,最终建立基于贝叶斯网络的变压器状态评估模型。实例验证了变压器状态评估模型的正确性和方案的可行性,基于贝叶斯网络的变压器状态评估模型能较好地满足工程需要,所提出的评估方法为变压器由定期预防性维修向状态维修的过渡提供了技术支持。

Condition Assessment for Power Transformers by Bayes Networks

Transformer condition assessment is the basis of performing condition-based maintenance of transformers.To obtain more accurate assessing results,a transformer is divided into three parts such as main body,bushing and core,accordingly,a new intelligent classifying model based on Bayesian networks is proposed to aim at the condition assessment of the above three parts of the transformer.During the assessing procedure,the future trend of key parameters related to a transformer part is firstly predicted according to the present and previous data of the parameters,then the historical,present and predictive conditions of the part are evaluated by considering all the scores of the key parameters,finally its comprehensive assessment can be obtained based on the proposed Bayesian network.To get the score of a transformer parameter,its threshold value and grade are obtained through a fuzzy membership function.To get the overall condition assessment of a transformer,a layered model is presented,in which the overall transformer condition is at the root,and the comprehensive conditions of the above three transformer parts are at the second layer.The proposed approach has been verified by the experimental data and condition samples of transformers of an electric utility company in China,and the results show that the proposed models have acceptable assessing ability.The proposed models are open and flexible,so the objective system is easy to develop and maintain,and the system can support condition-based maintenance for transformers.Moreover,the proposed condition assessment models based on Bayes networks can better meet the requirements of engineering.