smriti srivastava | University of Delhi (original) (raw)

Papers by smriti srivastava

Design of new neural networks is restricted due to some problems like stability, plasticity, comp... more Design of new neural networks is restricted due to some problems like stability, plasticity, computational complexity and memory consumption. These problems are overcome in the present work by using an interval feed-forward neural network (IFFNN). It has simple structure that reduces the computational complexity and memory consumption, and the use of Lyapunov stability (LS) based learning algorithm assures the stability. Effectiveness and applicability of the underlying IFFNN model is investigated on benchmark problems of identification.

Journal of Intelligent and Fuzzy Systems, 2013

For dealing with adjacent input fuzzy sets having overlapping information, non-additive fuzzy rul... more For dealing with adjacent input fuzzy sets having overlapping information, non-additive fuzzy rules are formulated by defining their consequent as a function of fuzzy measures, i.e., a simple form of Choquet integral. The fuzzy measures aggregate the information from the overlapping fuzzy sets using the λ-measure. The defuzzified output of these rules is also in the general form of the Choquet fuzzy integral. The underlying non-additive fuzzy model is investigated for both identification and control of non-linear systems. The identification of this fuzzy model involves the strength of the rules as the known input functions and fuzzy densities required to compute fuzzy measures as the unknown functions to be estimated. The use of q-measure provides a more flexible and powerful way of simplifying the computation of λ-measure used to take account of interaction between the fuzzy sets. This model has been successfully applied to the real life problem of verifying the authenticity of offline signatures.

$Research paper thumbnail of On comparing the performance of Genetic Algorithm optimized PI and fractional order PI controller for power quality enhancement$

This paper deals with the application of a optimization technique inspired by natural evolution, ... more This paper deals with the application of a optimization technique inspired by natural evolution, namely Genetic Algorithm (GA) for the design of Fractional order Proportional and Integral (FOPI) based DSTATCOM (Distributed Static Compensator) and ELC (Electronic Load Controller). The GA technique helps search efficiently the optimal parameters of the FOPI controller. Conventional controllers use integral order control which is less robust as compared to fractional order control. This paper is based on a novel application of fractional order controller optimized by genetic algorithm for power quality improvement using DSTATCOM and ELC in a power system. With the help of modelling carried out in a MATLAB based environment and a set of simulation results, the superiority of the designed FOPI over PI (Proportional — Integral) based DSTATCOM and ELC controllers used in power distribution system is affirmed and a comparative study in terms of overshoots and undershoots is presented.

Iete Journal of Research, 2007

ABSTRACT The present paper discusses an important issue of identification and control of a nonlin... more ABSTRACT The present paper discusses an important issue of identification and control of a nonlinear dynamical system using neural network. A novel method based on neural network model has been developed here which makes the task of calculating the system parameters simple in comparison to the methods reported so far for the extraction of system dynamics of a nonlinear system. The updated parameters of the system are obtained using a learning algorithm and the controller is tuned accordingly. Further the system stability is compared with that obtained by existing networks and the simulated results show that our method is superior.

This paper proposes a novel Wavelet Packet based feature extraction approach for the task of text... more This paper proposes a novel Wavelet Packet based feature extraction approach for the task of text independent speaker recognition. The features are extracted by using the combination of Mel Frequency Cepstral Coefficient (MFCC) and Wavelet Packet Transform (WPT).Hybrid Features technique uses the advantage of human ear simulation offered by MFCC combining it with multi-resolution property and noise robustness of WPT. To check the validity of the proposed approach for the text independent speaker identification and verification we have used the Gaussian Mixture Model (GMM) and Hidden Markov Model (HMM) respectively as the classifiers. The proposed paradigm is tested on voxforge speech corpus and CSTR US KED Timit database. The paradigm is also evaluated after adding standard noise signal at different level of SNRs for evaluating the noise robustness. Experimental results show that better results are achieved for the tasks of both speaker identification as well as speaker verification.

Environmental Technology & Innovation, 2021

Abstract With technological advancement taking place in almost every corner of the world, people ... more Abstract With technological advancement taking place in almost every corner of the world, people are just enjoying the increased comfort and luxury which this development is bringing with it. But an important issue which most of the people are overlooking is that this over-rushed development is bringing along with it a necessary evil — Pollution. Pollution in any form (air, water or land) is affecting almost all the life forms on the earth so rapidly that its adverse effects are being seen more prominently with each passing day. This paper throws light on two major classifications of air pollution i.e.outdoor and indoor pollution. It discusses their impact, sources and also possible ways to reduce their growing levels. The paper focuses on the analysis of pollution data from government source and predicts its possible future level using artificial intelligent tools. Modeling and prediction of carbon monoxide(CO) level in the atmosphere is performed taking vehicular emission as the major source of outdoor pollution. Fuzzy type-1 and type-2 systems have been used to model and predict the level of carbon monoxide in 3 metropolitan cities of India namely Bangalore, Delhi and Mumbai. Also a comparative analysis between the two methods is carried out and the simulation results clearly depict the superiority of the Fuzzy type-2 model. The effect of varying the learning rate used for learning the parameters is also shown in the paper.

2016 IEEE 1st International Conference on Power Electronics, Intelligent Control and Energy Systems (ICPEICES), 2016

For online control of various dynamical systems, an adaptive artificial neural network (ANN) base... more For online control of various dynamical systems, an adaptive artificial neural network (ANN) based proportional integral derivative (PID) controller is developed. For linear time invariant processes, conventional PID controller is suitable but they have limitations when they are required to control the plants having high non linearity or their parameters are changing with the time. In order to find the parameters of PID controller, information regarding the dynamics of the plant is essential. If perturbation occurs in plant parameter(s) then PID controller may work only if these changes are not severe. But most plants are either non linear or their parameters changes with time and this demands for a use of more robust type of controller and ANN is a suitable candidate. To use the power of PID controller and ANN, ANN based PID controller is proposed in this paper. The benefit of this combination is that it utilizes the simplicity of PID controller mathematical formula and uses the ANN powerful capability to handle parameter variations and non linearity.

International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, 2020

In a power distribution system, faults occurring can cause voltage sag that can affect critical l... more In a power distribution system, faults occurring can cause voltage sag that can affect critical loads connected in the power network which can cause serious effects in the oil and gas industry. The objective of this paper is to design and implement an efficient and economical dynamic voltage restorer (DVR) to compensate for voltage sag conditions in the oil and gas industry. Due to the complexity and sensitivity of loads, a short voltage sag duration can still cause severe power quality problems to the entire system. Dynamic Voltage Restorer (DVR) is a static series compensating type custom power device. The overall efficiency of the DVR largely relies on the effectiveness of the control strategy governing the switching of the inverters. It can be said that the heart of the DVR control strategy is the derivation of reference currents. This paper deals with the extraction of reference current values using a controller based on a combination of probabilistic and fuzzy set theory. The ...

Advances in Intelligent Systems and Computing, 2016

In this work, a speaker identification system is employed using mel-frequency cepstral coefficien... more In this work, a speaker identification system is employed using mel-frequency cepstral coefficients (MFCC) and gammatone frequency cepstral coefficients (GFCC) features. MFCC is the most common feature extraction technique used in speaker identification/verification system and gives high performance in clean environmental conditions. GFCC is known for its noise robustness performance and is highly suitable in noisy or office environment conditions. Here, we combine the advantages of both the feature extraction techniques by score level fusion. Also, we employed a more simpler Gaussian Membership Function (GMF) based matching process. Lastly, we use k-Nearest Neighbor (KNN) to measure the similarity in the verification stage. Experimental results verify the validity of our proposed approaches in personal authentication.

Applied Energy, 1995

Daily and hourly diffuse radiations measured using a pyranometer and shading ring have been compa... more Daily and hourly diffuse radiations measured using a pyranometer and shading ring have been compared with the values calculated using several correlations. For the daily diffuse radiation, the correlation of Duffie &amp;amp;amp;amp;amp;amp; Beckman (Solar Engineering of Thermal Processes, John Wiley, 1980), Liu &amp;amp;amp;amp;amp;amp; Jordan (Solar Energy, 4(3) (1960) 1–19), Erts et al., (Solar Energy, 28 (1982) 293) and Bruno (Solar

Abstract: The present paper discusses an important issue of control engineering Le. the extractio... more Abstract: The present paper discusses an important issue of control engineering Le. the extraction of system dynamics of a nonlinear system. For this a novel method based on neural network model has been developed which makes the task of calculating the system parameters ...

American Journal of Physics

We share the design for a simple apparatus that, when paired with an Arduino processor and a comp... more We share the design for a simple apparatus that, when paired with an Arduino processor and a computer, can be used in a wide range of laboratory measurements: observing linear kinematics, confirming Faraday's and Lenz's laws, measuring magnetic moments, and observing the effects of eddy currents. The setup is simple, inexpensive, easy to replicate, and can even be fabricated and used by students working at home.

Proceedings of the 2018 the 2nd International Conference on Video and Image Processing

Edge detection is a field of signal processing where the signal is an image. Edges contain most o... more Edge detection is a field of signal processing where the signal is an image. Edges contain most of the information making edge detection a very important image segmentation technique. Traditional edge detection uses integral order differentiation, the results of which are not satisfactory. Often, the edges are missing, fragmented or with false edges. In our proposed work, edge detection has been performed using fractional order calculus to overcome these drawbacks. Edges and noise, both are high frequency components and the presence of noise in an image can lead to erroneous results. Therefore, image denoising is performed before edge detection operation for a noisy test image.

Advances in Intelligent Systems and Computing, 2018

With the abundant demand for power and increasing number of loads causes a number of power qualit... more With the abundant demand for power and increasing number of loads causes a number of power quality disturbances. The detection and classification of these PQ disturbances have become a pressing concern. This paper presents a novel technique for disturbances classification in power distribution line using empirical mode decomposition of raw data and multi-layer perceptron method for classification. The electrical distribution model is designed over MATLAB/Simulink environment to create PQ disturbances. This proposed method successfully classifies five types of PQ disturbances, i.e., sag, swell, interruption, transients, harmonics, and one healthy for comparison, and obtains 98.9% correct classification rate for tested events.

This paper presents a comparative study of neural observers for TRMS. These are Chebyshev neural ... more This paper presents a comparative study of neural observers for TRMS. These are Chebyshev neural network based observer (CNN) and Multi-layer feedforward neural network (MLFFNN) observer. TRMS is a highly non-linear system having mutual interference between two rotors, it is trivial to design an effective controller for the TRMS to reach the desirable yaw and pitch angles. All the states are not available for the measurement, so to estimate the inaccessible states of TRMS these non-linear state observers are designed. On Comparing Performance MLFFNN found to be better than CNN observer. KeywordsTwin rotor multi input multi output system (TRMS), Chebyshev neural network (CNN) and Multi-layer feed-forward neural network (MLFFNN), Degree of freedom (DOF). Introduction Many real physical systems are nonlinear in nature. Controlling nonlinear systems is a difficult problem due to their complex nature. This problem becomes more acute when the system’s parameters are uncertain. Uncertainty...

2021 International Conference on Computing, Communication, and Intelligent Systems (ICCCIS), 2021

Direct marketing involves telephone calls, personalized emails, and messages, newsletters, which ... more Direct marketing involves telephone calls, personalized emails, and messages, newsletters, which catch the eye of the customer, and in turn, attract them towards the company. In this paper data of a Portuguese bank is considered. The bank makes a call to its potential customers regarding its term deposit schemes and is interested in customers that will invest in term deposits. With the advent of machine learning algorithms in prediction, comparative study of various algorithms such as Support Vector Machine (SVM), Gaussian Naïve Bayes, Random Forest, Light Gradient Boosting (GBM), and Extreme Gradient Boosting is performed for the said data set. Light GBM gives the best results among these in very less processing time.

International Journal of Adaptive Control and Signal Processing, 2018

In this paper, the problem of simultaneous identification and predictive control of nonlinear dyn... more In this paper, the problem of simultaneous identification and predictive control of nonlinear dynamical systems using self-recurrent wavelet neural network (SRWNN) is addressed. The structure of the SRWNN is a modification of the wavelet neural network (WNN). Unlike WNN, the neurons present in the hidden layer of SRWNN contain the weighted self-feedback loops. Dynamic back-propagation algorithm is employed to derive the necessary parameter update equations. To further improve the convergence speed of the parameters, a time-varying (adaptive) learning rate is used. Four simulation examples are considered for testing the effectiveness of the proposed method. Furthermore, some disturbance rejection tests are also performed on the proposed method. The results obtained through the simulation study confirm the effectiveness of the proposed method.

Journal of Intelligent & Fuzzy Systems, 2018

The present work describes different research techniques for collecting and organizing speech dat... more The present work describes different research techniques for collecting and organizing speech database in different scenario at the institute and successfully structuring the text independent speaker identification database in Indian context. In order to get the Multi-Scenario dataset, each speaker performed multiple sessions recording in reading style with English and Hindi language with same passages but under different conditions. This work analyzed different scenario affecting the performance of speaker recognition system when tested under dissimilar training conditions. Here four different scenarios are considered; sensor and environment, language, aging and health. To study the effect of sensor, language and environment on the performance of ASR system a database of 200 speaker was created. Under different environmental conditions, four different types of sensors in parallel configuration were used to study the sensor mismatch conditions over testing and training phase. The database contains speech samples of the individual in English and Hindi in read speech styles under two environment i.e. a controlled recording chamber and library. To study the aging effect, an aging NSIT speaker database (AG-NSIT-SD) of 53 famous personalities was collected from online source varying over a period of 10-20 years. Further to study the effect of health, a cough and cold NSIT speaker database (CC-NSIT-SD) of 38 speakers was also collected to study the performance of system. Apart from this, the effect of different noise types on the speaker identification was also studied on different sensors.

Soft Computing, 2019

In this paper, a noise-robust palmprint recognition system is discussed with a novel feature extr... more In this paper, a noise-robust palmprint recognition system is discussed with a novel feature extraction technique: twodimensional Cochlear transform (2D-CT) based on the textural analysis of image sample. Orthogonality of 2D-CT is proved which shows the high robustness of the proposed 2D-CT to noise. To validate the proposed feature extraction technique, palmprint recognition has been tested on both left and right palm of IITD database of 230 persons, CASIA palmprint database of 312 persons, polyU palmprint database of 386 persons and achieved high accuracy. The proposed 2D-CT method is compared with discriminative and robust competitive code, double orientation code, competitive coding, ordinal coding, Gabor transform, Gaussian membership-based features, absolute average deviation and mean features. Further, K-nearest neighbor is used to validate the matching stage. The results show superiority of the proposed method over other feature extraction methods.

Arabian Journal for Science and Engineering, 2017

Here, a highly precise human identification system is developed using a newly proposed biometric ... more Here, a highly precise human identification system is developed using a newly proposed biometric traitpalm-phalanges print (PPP). For PPP, NSIT database has been used which includes palm-phalanges. This database consists of anterior hand images of fifty individuals with ten samples each. To crop the region of interest from hand samples to get palm-phalanges, database is preprocessed. First, it has been shown that each finger phalange can be used as a biometric modality and give moderate/sufficient performance for low-accuracy system. For feature extraction, histograms of oriented gradients, GMF feature, mean and AAD methods have been used. To further enhance the performance, score-level and feature-level fusion strategies are applied and compared. Score-level fusion is performed using different fusion rules. Next for feature-level fusion, five methods are used: (1) simple concatenate, (2) PCA feature fusion, (3) linear discriminant analysis feature fusion, (4) fusion codes and (5) supervised local-preserving canonical correlation analysis method. Receiver operating characteristics (ROC), equal error rate, area under the curve of ROC and decidability index (d) are used to show the performance of the system qualitatively and quantitatively.