Measuring partial discharges in DC insulation systems is an issue due to the lack of a reference relating the voltage waveform to the physics of discharge phenomena. Also, DC is not always steady state, due to voltage and load transients that generate electric field profile variations inside an insulation system, which can affect partial discharge inception likelihood and characteristics. Partial discharge measurement technology must be able to separate discharge pulses from noise and identify the type of sources generating partial discharge, which is related to condition assessment and maintenance. Eventually, measurement and analysis should be automatic and unsupervised, in order to get rid, partially or totally, of expert support. This study addresses a new approach to partial discharge measurements in DC insulation systems, presenting algorithms for separation, recognition and identification, which are effective both in DC steady state and during voltage (and load) transients. These algorithms are automatic and do not require expert support. Various cases of algorithm application on test objects consisting of multilayer polymeric specimens with an internal cavity and defective cable models are presented and discussed. Their effectiveness is proved, at least at a laboratory level, with effective noise separation and identification of discharge typology.

中文翻译：

---

一种在电压瞬变和稳态期间直流绝缘系统局部放电测量和分析的创新方法

由于缺乏将电压波形与放电现象物理相关的参考，因此测量直流绝缘系统中的局部放电是一个问题。此外，直流并不总是稳定状态，因为电压和负载瞬变会在绝缘系统内产生电场分布变化，这会影响局部放电开始的可能性和特性。局部放电测量技术必须能够从噪声中分离出放电脉冲并识别产生局部放电的源类型，这与状态评估和维护有关。最终，测量和分析应该是自动的和无人监督的，以便部分或完全摆脱专家的支持。本研究提出了一种在直流绝缘系统中进行局部放电测量的新方法，提出用于分离、识别和识别的算法，这些算法在直流稳态和电压（和负载）瞬变期间均有效。这些算法是自动的，不需要专家支持。介绍并讨论了算法在测试对象上的各种应用案例，这些测试对象由具有内腔的多层聚合物样本和有缺陷的电缆模型组成。它们的有效性至少在实验室级别得到了证明，具有有效的噪声分离和排放类型识别。介绍并讨论了算法在测试对象上的各种应用案例，这些测试对象由具有内腔的多层聚合物样本和有缺陷的电缆模型组成。它们的有效性至少在实验室级别得到了证明，具有有效的噪声分离和排放类型识别。介绍并讨论了算法在测试对象上的各种应用案例，这些测试对象由具有内腔的多层聚合物样本和有缺陷的电缆模型组成。它们的有效性至少在实验室级别得到了证明，具有有效的噪声分离和排放类型识别。