【作者】 王永强；

【导师】 李和明； 律方成；

【作者基本信息】 华北电力大学（河北） ， 高电压与绝缘技术， 2009， 博士

【摘要】 电容型设备绝缘在线监测与故障诊断技术是实现其状态维修的前提和基础。电容型设备绝缘在线监测会受到环境因素(温度、湿度、污秽等)的影响,研究考虑环境影响的电容型设备绝缘监测与故障诊断技术具有理论意义和实用价值。本文在环境因素对电容型设备介质损耗因数(tanδ)影响的试验方法、tanδ的修正方法、电容型设备故障诊断方法等方面进行了深入的研究,并开发了在线监测与故障诊断系统。完成的主要工作和取得的主要研究成果如下:提出了全绝缘型人工气候室设计方案,计算分析了气候室内悬挂绝缘子和套管时的空间电场分布情况,并和金属材质气候室相比较,论证了该方案的优越性,据此方案研制的全绝缘人工气候室可调节温度、湿度和压力等参数,满足了人工气候试验的要求;开发了基于虚拟仪器技术的高压电容型设备绝缘特性自动测试系统,完成了人工气候室内变压器套管tanδ的在线测量试验,以此为基础分析了环境因素对电容型设备tanδ的影响情况,获得了tanδ与环境参数的关系曲线,建立了环境因素对tanδ影响的灰关联分析模型,得到了主要环境参数与电容型设备tanδ的灰关联度。提出了基于最小二乘支持向量机的主要环境参数对tanδ影响的修正方法,将不同气候环境条件下测得的电容型设备tanδ修正为标准大气条件下的值,从而使不同气候条件下的监测数据及离线试验数据具有了等效性,采用遗传算法优化了支持向量机参数,有效提高了修正精度,实例修正结果和实际的在线监测数据修正结果均验证了该方法的有效性。结合电容型设备的其他检测数据,提出了基于贝叶斯网络的电容型设备故障诊断方法,首次将贝叶斯网络方法引入到电容型设备的故障诊断中,并根据电容型设备故障诊断的特点改进了贝叶斯网络的推理过程,提高了该方法的实用性,诊断实例验证了该方法的有效性。通过实例分析验证了将环境参数与tanδ的灰关联度作为特征量进行电容型设备故障诊断的可行性,提出了基于灰关联度物元模型的电容型设备可拓诊断方法,该方法即排除了环境因素对tanδ的影响,又可以进行准确的故障分类,诊断实例验证了该方法的有效性。根据分布式电容型设备在线监测的需要,提出了多下位机同步采样控制方法,并开发了一套分层分布式电容型设备绝缘在线监测与故障诊断系统应用于实际工程,该系统采用就地数字化技术提高了抗干扰能力,经过测试与运行,系统运行可靠稳定,数据采集准确。更多还原

【Abstract】 To monitor and diagnose the insulation condition of capacitive equipments on-line is the precondition and basis to realize the condition-based maintenance (CBM) for capacitive equipments. On-line monitoring and fault diagnosis of capacitive equipment’s insulation would be influenced by environmental factors (such as temperature , humidity , contamination etc.). Therefore, it is very useful and meaningful to study the insulation monitoring and diagnosis method of capacitive equipments considering influence of environment. The test method of dielectric dissipation factor(tanδ)changing law in different environments, the tanδ’s modified method and the method of capacitive equipment’s fault diagnosis were researched in this paper. At the same time, on-line monitoring and fault diagnosis system was developed. The main work accomplished includes the following.In this paper, a new design proposal of fully insulated artificial climate chamber was put forward. The distribution of electric field in full space when insulators or bushing was hung in the chamber was analyzed. It was found that electrical field strength and potential around the the sample changed less in the fully insulated chamber than in conducting material phytotron, so it is closer the real condition. Based on the design, the first fully insulated large artificial climate chamber in China, in which temperature , humidity, and atmospheric pressure can be regulated, has been built in our laboratory. The paper developed an insulation parameters automatic test system of high voltage capacitive equipment based on LabVIEW. In the chamber, On-line tanδmeasurements of the capacitive transformer bushings were completed. The effects of environmental factors on tanδwere analyzed, and the curves between the tanδand the environment parameters were acquired. Then, a grey relational model to analyze the effects of main environmental factors on tanδwas built, and the grey relational grade between them was gained in this paper.In this paper, a modified model of the effects of main environmental factors on tanδbased on Least Squares Support Vector Machine( LS-SVM ) was put forward. So the tanδvalue measured in different climate conditions could be modified into a tanδvalue in standard atmospheric conditions, thereby the on-line measured data and the test data in laboratory under different climate conditions were equivalent. The model parameters were optimized by genetic algorithm, which improved the accuracy of modified calculation. The effectiveness of this method was validated by modified results of the test data and on-line monitoring data.Bayesian network method for capacitive equipment’s fault Diagnosis was presented by means of combining all kinds of monitor data based on Bayesian network theory. It is the first time that the paper introduced the Bayesian network theory into capacitive equipment’s fault diagnosis. Bayesian network’s reasoning process was improved based on the characteristics of capacitive equipment’s fault diagnosis in this paper. This method was proved to be effective by experiment.The feasibility that using the grey relational grade of main environmental factors on tanδas new characteristic quantity to diagnose the capacitive equipments was proved by examples. And the extension diagnosis method of capacitive equipment’s faults based on matter-element model was advanced. This method can not only exclude the effects of environmental factors on tanδ, but also realize fault classification exactly. The method was proved to be successful by experiment.A new control method of the synchronous data sampling on multi-lower computer system was developed according to the needs of distributed capacitive equipment on-line monitor. Base on this method, we developed an on-line insulation monitoring and fault diagnosis system for capacitive equipment, which was then applied in practice. This system took the locally digitizing process to improve anti-disturbing capability. In practical project , the operation of the system was stable and reliable, and the data sampling was accurate.更多还原