Introduction to statistical pattern recognition (2nd ed.)
Introduction to statistical pattern recognition (2nd ed.)
Feature Selection: Evaluation, Application, and Small Sample Performance
IEEE Transactions on Pattern Analysis and Machine Intelligence
The use of features selection and nearest neighbors rule for faults diagnostic in induction motors
Engineering Applications of Artificial Intelligence
Pattern recognition using discriminative feature extraction
IEEE Transactions on Signal Processing
Dimensionality reduction using genetic algorithms
IEEE Transactions on Evolutionary Computation
| Hi-index | 0.01 |
A pattern recognition technique associated with a new state estimator is developed in order to supervise electrical process. For this purpose, diagnostic features are extracted from current and voltage measurements for monitoring different operating modes. Then, a feature selection method is applied in order to select the most relevant features which define the feature space. In this frame, the classification is realized by a non-parametric method (''k-nearest neighbors'' rule) with reject options. However, this method does not take into account the evolution of the operating modes. Thus, it is necessary to enhance the initial knowledge database. For that, a polynomial approach allows characterizing the intermediate states of each operating modes and an original use of Kalman algorithm allows predicting the evolution of the partially known modes. A simple behavioral model is used to describe the evolution of the pattern vector. An estimation step provides the parameter of such model and a prediction step determines the future evolution of the pattern vector. This approach is illustrated on an asynchronous motor of 5.5kW, in order to detect broken bars under any load level. The experimental results prove the efficiency of pattern recognition methods in condition monitoring of electrical machines.