Informatics Publishing Limited

S. Himavathi ¹, A. Muthuramalingam ²

Affiliations
1 Electrical and Electronics Engineering Department, Pondicherry Engineering College, Puducherry-605014, IN
2 Electrical and Electronics Engineering Department, Pondicherry Engineering College, Puducherry-605014, IN

Abstract

Stator and rotor resistance variations occur during normal operation of Induction motors. The performance of a vector-controlled drive to a large extent depends on the accuracy of estimated flux, which in turn depends on estimation of rotor and stator resistance. This paper presents a neural based online method for estimation of stator resistance for indirect vector controlled drives. The neural based estimator is implemented using FPGA and tested using Icv400hq240-5. The results obtained are presented.

Keywords

Stator Resistance Estimators, Neural Based Estimators, FPGA Implementation.

Full Text

     

R. Subha ¹, S. Himavathi ²

Affiliations
1 Department of Electrical and Electronics Engineering, Sir M Visvesvaraya Institute of Technology, Bangalore – 562157, Karnataka, IN
2 Department of Electrical and Electronics Engineering, Pondicherry Engineering College, Puducherry – 605012, IN

Abstract

Background/Objectives: PV array being shaded partially by buildings, trees or passing clouds is common. This makes the P-V curve of the PV system complex with more than one peak. MPPT algorithm capable of consistently detecting the global peak within a short duration of time is essential. Methods/Statistical Analysis: Lately Particle Swarm Optimization (PSO) algorithm has been used for Maximum Power Point (MPP) tracking due to its ability to locate the MPP irrespective of its location in the P-V curve. This paper evaluates and compares the performance of the basic PSO algorithm and the modified PSO algorithms for ten different shading patterns. Findings: The basic PSO algorithm is compared with three modified PSO algorithms - PSO algorithm with random numbers eliminated, PSO algorithm with linearly varying constants and PSO algorithm with fixed maximum iterations. The basic PSO algorithm gives good results but random numbers in the algorithm tends to make the convergence time random for the same shading pattern and makes hardware implementation difficult. The PSO algorithm with random numbers eliminated overcomes this disadvantage and is found to give good results. But the convergence time is a little higher and varies with shading pattern. The PSO algorithm with fixed maximum iterations gives good performance with shorter and fixed convergence time. Application/Improvements: PSO algorithm with fixed maximum iterations thus improves the responsiveness of the algorithm to rapidly changing patterns of shading.

Keywords

Maximum Power Point Tracking, Partial Shading, Particle Swarm Optimization, PV Array.

Full Text

   

S. Himavathi ¹, A. Muthuramalingam ¹, A. Venkadesan ¹, K. Sedhuraman ¹

Affiliations
1 Department of Electrical and Electronics Engineering, Pondicherry Engineering College, IN

Abstract

Neural Networks (NN) have proved its efficacy for nonlinear system modeling. NN based controllers and estimators for nonlinear systems provide promising alternatives to the conventional counterpart. However, NN models have to meet the stringent requirements on execution time for its effective use in real time applications. This requires the NN model to be structurally compact and computationally less complex. In this paper a parametric method of analysis is adopted to determine the compact and faster NN model among various neural network architectures. This work proves through analysis and examples that the Single Neuron Cascaded (SNC) architecture is distinct in providing compact and simpler models requiring lower execution time. The unique structural growth of SNC architecture enables automation in design. The SNC Network is shown to combine the advantages of both single and multilayer neural network architectures. Extensive analysis on selected architectures and their models for four benchmark nonlinear theoretical plants and a practical application are tested. A performance comparison of the NN models is presented to demonstrate the superiority of the single neuron cascaded architecture for online real time applications.

Keywords

Single Neuron Cascading, Neural Networks, Modeling, Compact Models, Real-time Applications.

Full Text

     

J. Pradeep ¹, E. Srinivasan ¹, S. Himavathi ²

Affiliations
1 Department of ECE, Pondicherry College Engineering, Pondicherry, IN
2 Department of EEE, Pondicherry College Engineering, Pondicherry, IN

Abstract

An off-line handwritten alphabetical character recognition system using multilayer feed forward neural network is described in the paper. A new method, called, diagonal based feature extraction is introduced for extracting the features of the handwritten alphabets. Fifty data sets, each containing 26 alphabets written by various people, are used for training the neural network and 570 different handwritten alphabetical characters are used for testing. The proposed recognition system performs quite well yielding higher levels of recognition accuracy compared to the systems employing the conventional horizontal and vertical methods of feature extraction. This system will be suitable for converting handwritten documents into structural text form and recognizing handwritten names.

Keywords

Handwritten Character Recognition, Image Processing, Feature Extraction, Feed forward Neural Networks.

Full Text