Dissolved gas analysis (DGA) is attracting greater and greater interest from researchers as a fault diagnostic tool for power transformers filled with vegetable insulating oils. This paper presents experimental results of dissolved gases in insulating oils under typical electrical and thermal faults in transformers. The tests covered three types of insulating oils, including two types of vegetable oil, which are camellia insulating oil, Envirotemp FR3, and a type of mineral insulating oil, to simulate thermal faults in oils from 90˝C to 800˝C and electrical faults including breakdown and partial discharges in oils. The experimental results reveal that the content and proportion of dissolved gases in different types of insulating oils under the same fault condition are different, especially under thermal faults due to the obvious differences of their chemical compositions. Four different classic diagnosis methods were applied: ratio method, graphic method, and Duval's triangle and Duval's pentagon method. These confirmed that the diagnosis methods developed for mineral oil were not fully appropriate for diagnosis of electrical and thermal faults in vegetable insulating oils and needs some modification. Therefore, some modification aiming at different types of vegetable oils based on Duval Triangle 3 were proposed in this paper and obtained a good diagnostic result. Furthermore, gas formation mechanisms of different types of vegetable insulating oils under thermal stress are interpreted by means of unimolecular pyrolysis simulation and reaction enthalpies calculation.

No abstract

An environment-friendly insulating gas, perfluoroisobutyronitrile (C4F7N), has been developed recent years. Due to its relatively high liquefaction temperature (around −4.7 °C), buffer gases, such as CO2 and N2, are usually mixed with C4F7N to increase the pressure of the filled insulating medium. During these processes, the insulating gases may be contaminated with micro-water, and the mixture of H2O with C4F7N could produce HF under breakdown voltage condition, which is harmful to the gas insulated electricity transfer equipment. Therefore, removal of H2O and HF in situ from the gas insulated electricity transfer equipment is significant to its operation security. The adsorbents with the ability to remove H2O but without obvious C4F7N/CO2 adsorption capacity are essential to be used in this system. In this work, a series of industrial adsorbents and desiccants were tested for their compatibility with C4F7N/CO2. Pulse adsorption tests were conducted to evaluate the adsorption performance of these adsorbents and desiccants on C4F7N and CO2. The 5A molecular sieve showed high adsorption of C4F7N (22.82 mL/g) and CO2 (43.86 mL/g); F-03 did not show adsorption capacity with C4F7N, however, it adsorbed CO2 (26.2 mL/g) clearly. Some other HF adsorbents, including NaF, CaF2, MgF2, Al(OH)3, and some desiccants including CaCl2, Na2SO4, MgSO4 were tested for their compatibility with C4F7N and CO2, and they showed negligible adsorption capacity on C4F7N and CO2. The results suggested that these adsorbents used in the gas insulated electricity transfer equipment filled with SF6 (mainly 5A and F-03 molecular sieves) are not suitable anymore. The results of this work suggest that it is a good strategy to use a mixture of desiccants and HF adsorbents as new adsorbents in the equipment filled with C4F7N/CO2.

No abstract

Aiming at the problem of similarity calculation error caused by the extremely sparse data in collaborative filtering recommendation algorithm, a collaborative filtering recommendation algorithm based on slope one matrix prefilling model, principal component dimension reduction, and binary K -means clustering is proposed in this paper. Firstly, the algorithm uses the slope one model based on item similarity to prefill the original scoring matrix. Secondly, principal component analysis is used to reduce the dimension of the filled matrix, retain the most representative dimension of user characteristics, and remove the dimension with less information. Finally, in order to solve the time-consuming problem of similarity calculation of collaborative filtering algorithm in the case of large-scale system, binary K -means clustering is carried out in the reduced dimension vector space to reduce the search range of the nearest neighbour of the target user. The algorithm ensures the efficiency and accuracy of recommendation while the scale of users is expanded. The experimental results on movielens dataset show that the algorithm proposed in this paper is superior to the traditional collaborative filtering algorithm and the collaborative filtering recommendation algorithm based on PCA (principal component analysis) and binary K -means clustering in recall rate, accuracy rate, average error, and running time.