1. |
- Axelberg, Peter G.V. 1959, et al.
(författare)
-
AUTOMATIC CLASSIFICATION OF VOLTAGE EVENTS USING THE SUPPORT VECTOR MACHINE METHOD
- 2007
-
Ingår i: 19th International Conference on Electricity Distribution (SIRED 2007) , Vienna, Austria, 21-24 May, 2007.
-
Konferensbidrag (refereegranskat)abstract
- Statistically based classification systems need to be trained on a large number of training data in order to classify unseen data accurately. However, it is difficult to gather enough voltage events for the training purpose from real recordings. Therefore, a classification system trained to accurately classify real voltage events, but based on synthetic training data is highly in demand. This paper therefore proposes the design of a statistically based classification system trained on synthetic data. The paper gives also the results of conducted performance tests when the proposed classification system was trained to classify seven common types of voltage events. The experiments showed an overall detection rate of 81.6%, 91.9% and 99.5% respectively.
|
|
2. |
- Axelberg, Peter G.V. 1959, et al.
(författare)
-
Support Vector Machine for Classification of Voltage Disturbances
- 2007
-
Ingår i: accepted for publication in IEEE Transactions on Power Delivery. ; 22:3, s. 1297-1303, July, 2007
-
Tidskriftsartikel (refereegranskat)abstract
- The Support Vector Machine (SVM) is a powerful method for statistical classification of data used in a number of different applications. However, the usefulness of the method in a commercial available system is very much dependent on whether the SVM classifier can be pre-trained from a factory since it is not realistic that the SVM classifier must be trained by the customers themselves before it can be used. We first propose a novel SVM classification system for voltage disturbances. Our aim also includes investigating the performance of the proposed SVM classifier when the voltage disturbance data used for training and testing are originated from different sources. The data used in the experiments were originated from both real disturbances recorded in two different power networks and from synthetic data. The experimental results have shown excellent accuracy in classification when training data were originated from one power network and unseen testing data from another. High accuracy was also achieved when the SVM classifier was trained on data from a real power network and test data originated from synthetic data. Slightly less accuracy was achieved when the SVM classifier was trained on synthetic data and test data were originated from the power network.
|
|
3. |
- Axelberg, Peter G.V. 1959, et al.
(författare)
-
Trace of flicker sources by using the quantity of flicker power
- 2007
-
Ingår i: IEEE transactions on Power Delivery. ; 23:1, s. pp.465-471
-
Tidskriftsartikel (refereegranskat)abstract
- Industries that produce flicker are often placed close to each other and connected to the same power grid system. This implies that the measured flicker level at the point of common coupling (PCC) is a result of contribution from a number of different flicker sources. In a mitigation process it is essential to know which one of the flicker sources is the dominant one. We propose a method to determine the flicker propagations and trace the flicker sources by using flicker power measurements. Flicker power is considered as a quantity containing both sign and magnitude. The sign determines if a flicker source is placed downstream or upstream with respect to a given monitoring point and the magnitude is used to determine the propagation of flicker power throughout the power network and to trace the dominant flicker source. This paper covers the theoretical background of flicker power and describes a novel method for calculation of flicker power that can be implemented in a power network analyzer. Also conducted simulations and a field test based on the proposed method will be described in the paper.
|
|