Support Vector Data Description Approach Research Articles

A machine learning approach for modeling irregular regions with multiple owners in wind farm layout design

Wind farm development projects require a detailed survey of the eligible land. The land selected is often segmented into different region, each owned by different landowners with different land pricing. These regions are often complex shaped with irregular boundaries. Therefore, an efficient method for numerically modeling such irregular domains is needed. This work uses support vector data description (SVDD) to generate an analytical, continuous description of the irregular regions. Whereas other methods typically work well for modeling convex domains, the SVDD approach can be used to model irregular regions as a spherical boundary using various kernel mapping. It was demonstrated that the SVDD approach can be used to model any number of complex regions. An error analysis showed that the SVDD approach can construct accurate descriptions using a relatively small data set. The applicability of SVDD method in wind farm layout optimization is also demonstrated. The wind farm optimization study considered that the terrain is divided into several regions each owned by a different owner offering the land at a different price. Two different methods for considering the cost of the land are presented. The differences in optimal farm layouts using the two land cost models were also presented. In each case, the optimized wind farm layouts resulted in lower cost-of-energy relative to the reference wind farm. It was shown that the SVDD approach can also be used to restrict the placement of wind turbines in infeasible/restricted regions. The library for support vector data description was also made available to the public.

Power transmission risk assessment considering component condition

This paper proposes a new method for power transmission risk assessment considering historical failure statistics of transmission systems and operation failure risks of system components. Component failure risks are integrated into the new method based on operational condition assessment of components using the support vector data description (SVDD) approach. The traditional outage probability model of transmission lines has been modified to build a new framework for power transmission system risk assessment. The proposed SVDD approach can provide a suitable mechanism to map component assessment grades to failure risks based on probabilistic behaviors of power system failures. Under the new method, both up-to-date component failure risks and traditional system risk indices can be processed with the proposed outage model. As a result, component failure probabilities are not only related to historical statistic data but also operational data of components, and derived risk indices can reflect current operational conditions of components. In simulation studies, the SVDD approach is employed to evaluate component conditions and link such conditions to failure rates using up-to-date component operational data, including both on-line and off-line data of components. The IEEE 24-bus RTS-1979 system is used to demonstrate that component operational conditions can greatly affect the overall transmission system failure risks.

A Support Vector Data Description Approach to NLOS Identification in UWB Positioning

Non-line-of-sight (NLOS) propagation is one of the most important challenges in radio positioning, and, in recent years, significant attention has been drawn to the identification and mitigation of NLOS signals. This paper focuses on the identification of NLOS signals. The authors consider the NLOS identification problem as a one-class classification problem and apply the support vector data description (SVDD), providing accurate data descriptions utilizing kernel techniques, to perform NLOS identification in ultrawide bandwidth (UWB) positioning. Our work is based on the fact that some features extracted from the received signal waveforms, such as the kurtosis, the mean excess delay spread, and the root mean square delay spread, are different between line-of-sight (LOS) and NLOS signals. Numerical simulations are performed to demonstrate the performance, using a dataset derived from a measurement campaign.

One-class classification with Gaussian processes

Detecting instances of unknown categories is an important task for a multitude of problems such as object recognition, event detection, and defect localization. This article investigates the use of Gaussian process (GP) priors for this area of research. Focusing on the task of one-class classification, we analyze different measures derived from GP regression and approximate GP classification. We also study important theoretical connections to other approaches and discuss their underlying assumptions. Experiments are performed using a large number of datasets and different image kernel functions. Our findings show that our approaches can outperform the well-known support vector data description approach indicating the high potential of Gaussian processes for one-class classification. Furthermore, we show the suitability of our methods in the area of attribute prediction, defect localization, bacteria recognition, and background subtraction. These applications and experiments highlight the easy applicability of our method as well as its state-of-the-art performance compared to established methods.

Improving the SVDD Approach to Hyperspectral Image Classification

In recent studies, the support vector data description (SVDD) has been successfully applied to the classification of hyperspectral images. However, there is a major problem with this approach, namely, the precise setting of the Gaussian kernel width (i.e., the sigma), which is, in fact, the common limitation of kernel methods in achieving a reliable performance. Generally, the sigma is tuned for multiclass data sets through the <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> -fold cross validation ( <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> CV), a time-consuming method. To reduce the computation time in real-time applications, typically, the <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> CV is used to constrain all the involved SVDD classifiers to share the same sigma. This letter presents a fast and straightforward method to estimate the sigma for each individual SVDD classifier based on statistical properties of the Gaussian kernel. To evaluate the performance of the proposed method, three frequently used hyperspectral data sets are employed. The results are then compared to the <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> CV method for sigma selection, and, in addition, two direct sigma estimation methods. Preliminary results using incomplete training data suggest that the proposed method can achieve similar or better performance with faster processing times than the <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> CV and also provide a significant superior performance in comparison with the direct methods.

A SVDD approach of fuzzy classification for analog circuit fault diagnosis with FWT as preprocessor

In this paper, a new approach of fault diagnosis in analog circuits, which employs the Fractional Wavelet Transform (FWT) to extract fault features and adopts a fuzzy multi-classifier based on the Support Vector Data Description (SVDD) to diagnose circuit faults, is proposed. Firstly, a discrete FWT algorithm by the fractional kernel matrix is performed to preprocess fault samples. To obtain the optimal fractional order, two methods trained with the genetic algorithm are introduced. One approach is performed by the best diagnostic result, and the other is based on the maximum among-cluster center distance by the Kernel Fuzzy C-Means (KFCM) algorithm. In this paper, a threshold value is used to decrease the fuzzy region which in the overlap between hyperspheres of SVDD. Then, a SVDD fuzzy multi-classifier is applied to diagnose faults in analog circuit, and fuzzy faults are diagnosed in fuzzy sets by the relative distance. The simulation results show that the FWT succeeds in extracting local fault features and the classifier effectively detects faults.