Fault detection and location are important and front-end tasks in assuring the reliability of power electronic circuits. In essence, both tasks can be considered as the classification problem. This paper presents a fast fault classification method for power electronic circuits by using the support vector machine (SVM) as a classifier and the wavelet transform as a feature extraction technique. Using one-against-rest SVM and one-against-one SVM are two general approaches to fault classification in power electronic circuits. However, these methods have a high computational complexity, therefore in this design we employ a directed acyclic graph (DAG) SVM to implement the fault classification. The DAG SVM is close to the one-against-one SVM regarding its classification performance, but it is much faster. Moreover, in the presented approach, the DAG SVM is improved by introducing the method of Knearest neighbours to reduce some computations, so that the classification time can be further reduced. A rectifier and an inverter are demonstrated to prove effectiveness of the presented design.

The paper presents an example of Instance-Based Learning using a supervised classification

method of predicting selected ductile cast iron castings defects. The test used the algorithm

of k-nearest neighbours, which was implemented in the authors’ computer application. To

ensure its proper work it is necessary to have historical data of casting parameter values

registered during casting processes in a foundry (mould sand, pouring process, chemical

composition) as well as the percentage share of defective castings (unrepairable casting defects).

The result of an algorithm is a report with five most possible scenarios in terms of

occurrence of a cast iron casting defects and their quantity and occurrence percentage in

the casts series. During the algorithm testing, weights were adjusted for independent variables

involved in the dependent variables learning process. The algorithms used to process

numerous data sets should be characterized by high efficiency, which should be a priority

when designing applications to be implemented in industry. As it turns out in the presented

mathematical instance-based learning, the best quality of fit occurs for specific values of

accepted weights (set #5) for number k = 5 nearest neighbours and taking into account the

search criterion according to “product index”.

One of the most important issues that power companies face when trying to reduce time and cost maintenance is condition monitoring. In electricity market worldwide, a significant amount of electrical energy is produced by synchronous machines. One type of these machines is brushless synchronous generators in which the rectifier bridge is mounted on rotating shafts. Since bridge terminals are not accessible in this type of generators, it is difficult to detect the possible faults on the rectifier bridge. Therefore, in this paper, a method is proposed to facilitate the rectifier fault detection. The proposed method is then evaluated by applying two conventional kinds of faults on rectifier bridges including one diode open-circuit and two diode open-circuit (one phase open-circuit of the armature winding in the auxiliary generator in experimental set). To extract suitable features for fault detection, the wavelet transform has been used on recorded audio signals. For classifying faulty and healthy states, K-Nearest Neighbours (KNN) supervised classification method was used. The results show a good accuracy of the proposed method.