中文摘要：

为减小变压器不满设计寿命便提前退役为电网带来的巨大损失,提出基于健康水平诊断技术的变压器综合寿命评估模型。该模型收集了目前可用于评估变压器寿命信息的特征参量,多源异构参量的融合可削弱单个参量引起的误差,提高寿命预测模型的精确度。将变压器状态量按层次分析结构分为基本参数、运行参数、可靠性参数、不良工况参数4大类,其中,第1层融合对主绝缘老化状态存在长期影响的运行环境与长期运行负荷,并以此修正由设计寿命计算得到的预期运行寿命;第2层考虑绝缘运行状态的设备个体差异,利用模糊推理、阈值诊断等来评估变压器的运行及试验情况,各个参量对寿命的影响程度依据层次结构的权重分配来体现;第3层考虑管理系统中同一地区变压器的部件缺陷风险情况,建立起符合威布尔分布的部件可靠性评价模型,求得部件运行可靠度;第4层提出运用热点温升模型对短路及严重过热等不良工况对模型进行修正。最后,结合实例分析变压器剩余寿命及故障概率,验证了模型的有效性,为变压器的状态检修、延长实际使用寿命提供参考依据。

英文摘要：

To decrease the massive loss due to the early retirement of power transformers, based on health index （HI） an integrated life estimation model is proposed. Since the merging of multi-source heterogeneous parameters can reduce the error caused by single parameter and increase the accuracy of longevity prediction model, the proposed model collects characteristic parameters, which can be used to assess the longevity information in the present. Based on hierarchical architecture the quantities of the power transformer state are reclassified into four categories, namely the fundamental parameters, the operational and maintenance parameters, the reliability parameters and the harmful operating condition parameters, and they belong to different layers respectively. In the first layer the operational environment and long-term operating load that impact the ageing condition of main insulation of power transformer in the long term are merged and on this basis the anticipated operating life obtained from the calculation of the designed lifetime by correction factors is revised; in the second layer the individual differences among insulation operating states of components are considered and the operating condition and test situation of power transformer are assessed by fuzzy reasoning and threshold diagnosis, and the impacting extent of each parameter on the lifetime is incarnated by the weight allocation of the hierarchical architecture; in the third layer the risk, which is caused by anticipated or running deficiencies of transformer components located at the same area, is considered in the management system, and a component reliability assessment model that conforms to Weibull distribution is established to get the operating reliability of components; in the fourth layer the hotspot temperature raising model is utilized to revise the established reliability model under harmful operating conditions such as short circuit and serious overheating. Finally, combined with the case the residual life and the fault probability