Birendra Biswal

Automatic Classification of Power Quality Events Using Balanced Neural Tree

This paper proposes an empirical-mode decomposition (EMD) and Hilbert transform (HT)-based method for the classification of power quality (PQ) events. Nonstationary power signal disturbance waveforms are considered as the superimposition of various undulating modes, and EMD is used to separate out these intrinsic modes known as intrinsic mode functions (IMFs). The HT is applied on all the IMFs to extract instantaneous amplitude and frequency components. This time-frequency analysis results in the clear visual detection, localization, and classification of the different power signal disturbances. The required feature vectors are extracted from the time-frequency distribution to perform the classification. A balanced neural tree is constructed to classify the power signal patterns. Finally, the proposed method is compared with an S-transform-based classifier to show the efficacy of the proposed technique in classifying the PQ disturbances.

Forecasting and classification of Indian stocks using different polynomial functional link artificial neural networks

Forecasting stock price index is one of the major challenges in the trade market for investors. Time series data for prediction are difficult to manipulate, but can be focused as segments to discover interesting patterns. In this paper we use several functional link artificial neural networks to get such patterns for predicting stock indices. The novel architecture of functional link artificial neural network with working principle of different models are provided to achieve best forecasting and classification with increase in accuracy of prediction and decrease in training time. Various FLANN models with different polynomials are investigated using different Indian stock indices like IBM, BSE, Oracle, & RIL. The main absolute percentage error (MAPE), sum squared error (SSE) and the standard deviation error (SDE) have been considered to measure the performance of the different FLANN models. In this paper we have presented the result using Reliance Industries Limited (RIL) stock data between 22/12/1999 to 30/12/2011 on closed price of every trading day.

Nonstationary power signal time series data classification using LVQ classifier

A new approach to time-frequency analysis and pattern recognition of non-stationary power signals is proposed in this paper. In this manuscript, visual localization, detection and classification of non-stationary power signals are achieved using wavelet packet decomposition and automatic pattern recognition is carried out through learning vector quantization neural network. The wavelet packet decomposition (WPD) of the non-stationary power signals is carried out to extract the coefficients at multiple level of decomposition. The relevant features for pattern classification are derived from the time-scale information obtained by WPD. The extracted features are used to classify different power quality disturbances by using learning vector quantization neural net. Various non-stationary power signal waveforms are considered to verify the applicability of the proposed technique.

Comparative study of stock market forecasting using different functional link artificial neural networks

This paper presents different forecasting functional link artificial neural network (FLANN) models to investigate and compare various time series stock data. The architecture of several FLANN models like CFLANN, LFLANN, LeF-LANN, and CEFLANN are discussed. The processing technique and experimental results are provided to investigate the prediction of stocks. This piece of work presents the training and testing of all the models by analysing and forecasting different Indian stocks like IBM, RIL and DWSG. All the forecasting models have been tested using same duration time of time series data. The experimental results illustrate that the trigonometric polynomial-based CEFLANN model outperforms the forecasting time series stock data in terms of percentage average error than the polynomial-based FLANN models. Lastly, the percentage of average error is further improved by optimising the free parameters of the trigonometric polynomial-based CEFLANN model with differential evolution algorithm (DEA).

Security of image in multimedia applications

In this paper, we propose an independent efficient compression and encryption techniques for an image in multimedia applications. In order to reduce the volume of multimedia data over wireless channels, data compression techniques are widely used. Discrete cosine transform (DCT) is one of the major compression Scheme. In this paper, we present Dynamic bit-width adaptation scheme in discrete cosine transform (DCT) as an efficient compression technique. We select the appropriate operand bit widths that achieve significant reduction of power consumption. It is essential to protect the confidential image data from unauthorized access for multimedia applications. Hence, in this paper, we present a modification to the Advanced Encryption Standard (MAES) to reflect a high level security and better image encryption. The modification is done by adjusting the Shift Row Transformation.

A hybrid FLANN and adaptive differential evolution model for forecasting of stock market indices

This paper presents a computationally efficient functional link artificial neural network CEFLANN based adaptive model for financial time series prediction of leading Indian stock market indices. Financial time-series data are usually non-stationary and volatile in nature. The proposed adaptive CEFLANN based model employs the least mean square LMS algorithm with a new cost function to train the weights of the networks. The mean absolute percentage error MAPE with respect to actual stock prices is selected as the performance index to estimate the quality of prediction. The CEFLANN model inputs are chosen from the past stock prices of different market sectors along with technical indicators to determine best stock trend prediction one day ahead in time. Further to improve the performance of the CEFLANN model, weights are optimized using an adaptive differential evolution DE algorithm and its overall prediction performance is compared with the improved LMS algorithm showing the effectiveness of the DE in producing more accurate forecast. We have selected different combinations of important technical indicators to have a strong control on changes in stock indices.

A New Approach to Time–Time Transform and Pattern Recognition of Non-stationary Signal Using Fuzzy Wavelet Neural Network

This paper discusses new approaches in time- time transform and Nonstationary power signals classification using fuzzy wavelet neural networks. The time-time representation is derived from the S-transform, a method of representation of a real time series as a set of complex, time-localized spectra. When integrated over time, the S-transform becomes the Fourier transform of the primary time series. Similarly, when summed over the primary time variable, the TT-transform reverts to the primary time series. TT-transform points to the possibility of filtering and signal to noise improvements in the time domain. In our research work visual localization, detection and classification of Nonstationary power signals problem using TT-transform and automatic Nonstationary power signal classification using FWNN (Fuzzy wavelet Neural Network) have been considered. Time-time analysis and Feature extraction from the Nonstationary power signals is done by TT-transform. In the proposed work pattern recognition of various Nonstationary power signals have been considered using particle swarm optimization technique. This paper also emphasizes the robustness of TT-transform towards noise. The average classification accuracy of the noisy signals due to disturbances in the power network is of the order 92.1.

Polynomial Based Functional Link Artificial Recurrent Neural Network adaptive System for predicting Indian Stocks

AbstractA low complexity Polynomial Functional link Artificial Recurrent Neural Network (PFLARNN) has been proposed for the prediction of financial time series data. Although different types of polynomial functions have been used for low complexity neural network architectures earlier for stock market prediction, a comparative study is needed to choose the optimal combinations of the nonlinear functions for a reasonably accurate forecast. Further a recurrent version of the Functional link neural network is used to model more accurately a chaotic time series like stock market indices with a lesser number of nonlinear basis functions. The proposed PFLARNN model when trained with the well known gradient descent algorithm produces reasonable accuracy with a choice of range of weight parameters of the network. However, to improve the accuracy of the forecast further, the weight parameters of the recurrent functional neural network are optimized using an evolutionary learning algorithm like the differential evo...

Detection and power quality disturbance classification based on wavelet packet decomposition and modified immune optimization algorithm

In this paper, a new approach to time-frequency analysis and pattern recognition of non-stationary power signals is proposed. In this paper, visual localization, detection and classification of non-stationary power signals are achieved using wavelet packet decomposition. Automatic pattern recognition is carried out through Modified Immune Optimization Algorithm MIOA based Reformulated Fuzzy C-means Algorithm. Time-frequency analysis and feature extraction from the non-stationary power signals are done by wavelet packet decomposition WPD. Various non-stationary power signal waveforms are processed through wavelet packet decomposition to generate time-frequency contours for extracting relevant features for pattern classification. The extracted features are clustered using Reformulated Fuzzy C-means Algorithm and finally the algorithm is extended using Artificial Immune and Modified Immune Optimization Algorithm MIOA respectively to refine the cluster centers. Results of simulation and analysis demonstrate that the proposed MIOA method achieves higher classification rate, better convergence property.

Dynamic Recurrent FLANN Based Adaptive Model for Forecasting of Stock Indices

Prediction of future trends in financial time-series data are very important for decision making in the share market. Usually financial time-series data are non-linear, volatile and subject to many other factors like local or global issues, causes a difficult task to predict them consistently and efficiently. This paper present an improved Dynamic Recurrent FLANN (DRFLANN) based adaptive model for forecasting the stock Indices of Indian stock market. The proposed DRFLANN based model employs the least mean square (LMS) algorithm to train the weights of the networks. The Mean Absolute Percentage Error (MAPE), the Average Mean Absolute Percentage Error (AMAPE), the variance of forecast errors (VFE) is used for determining the accuracy of the model. To improve further the forecasting results, we have introduced three technical indicators named Relative Strength Indicator (RSI), Price Volume Change Indicator (PVCI), and Moving Average Volume Indicator (MAVI). The reason of choosing these three indicators is that they are focused on important attributes of price, volume, and combination of both price and volume of stock data. The results show the potential of the model as a tool for making stock price prediction.