Chetan B. Khadse

Bio: Chetan B. Khadse is an academic researcher from Massachusetts Institute of Technology. The author has contributed to research in topics: Artificial neural network & Backpropagation. The author has an hindex of 3, co-authored 9 publications receiving 82 citations. Previous affiliations of Chetan B. Khadse include Visvesvaraya National Institute of Technology & Indian Institutes of Information Technology.

Electromagnetic Compatibility Estimator Using Scaled Conjugate Gradient Backpropagation Based Artificial Neural Network

Abstract: In this paper, an electromagnetic compatibility estimator is proposed using an artificial neural network with a scaled conjugate gradient algorithm. Neural networks are trained with the help of seven different optimization algorithms in MATLAB. Their performance is evaluated on the basis of number of neurons, desired output, and mean-squared error in offline mode in MATLAB. Among seven algorithms, scaled conjugate gradient algorithm is found to be the best choice. Hence, it is implemented in LabVIEW for online assessment of electromagnetic compatibility issues. Voltage dip, swell, and harmonics are generated with the help of an experimental setup. It consists of 230 V, 50 Hz input voltage supply, microcontroller, variac, and solid-state relays. It is interfaced to the LabVIEW software with the help of an NI USB 6361 data acquisition system. It enabled the continuous online monitoring of various signals. Along with voltage dip and swell, harmonics are also evaluated with the help of spectrum analyzer in LabVIEW. The detailed description of a hardware setup and mathematical modeling of trained network is given in this paper.

Comparison of seven backpropagation algorithms for three phase power quality assessment

Abstract: In this paper, comparison of seven backpropagation algorithms used in neural network is made for the three phase power quality assessment. Voltage sag and swell are the two disturbances taken into account for comparing the algorithms. These disturbances are generated with the help of programming in MATLAB. The input data, to train the network is the generated sag and swell disturbances. The backpropagation algorithms are taken into account for comparison are conjugate gradient descent, Levenberg-Marquardt, one step secant, Baysian regularization, scaled conjugate gradient descent, gradient descent with momentum and gradient descent with adaptive learning rate. The seven algorithms are used to train the seven networks and tested using simulation model in MATLAB. After testing is done, the comparison is made on the basis of mean squared error, number of neurons, percent of accurate cases detected, number of iterations required for the training process. The testing results for the each algorithm and comparison are presented.

A laboratory set-up for power quality disturbance generator and real time power quality monitoring

Abstract: Solid State relay based single phase power quality disturbance generator is proposed in this paper. Digital control system based on microcontroller 89C51 is implemented to control the duration of power quality disturbances. Voltage sag, swell and interruption are generated. Dimmerstat is used to change the voltage levels and so to control the intensity of disturbances. Design of proposed generator is discussed. Simple design and use of easily available components makes this generator economic. It is highly suitable in power quality laboratory for experimental work. Students working on voltage sag/swell compensation and monitoring of power quality events can use this generator. Proposed generator is used as an application for PC based real time power quality monitoring system. Monitoring system is designed in LabVIEW. For fetching the real time signal, voltage sensor and data acquisition system are used. Results of real time detection of disturbances and classification of disturbances into momentary, temporary, instantaneous and under voltage or over voltage are done by monitoring system and presented in this paper.

Artificial Neural Network based Fault Detection System for 11 kV Transmission Line

Abstract: In this paper, a fault identification system based on the artificial neural network is proposed. An 11 KV transmission line model is developed in MATLAB simulink model. The faults under considerations are line to ground, line to line, double line to ground, triple line and triple line to ground. These faults are created with the help of fault creation block. The duration as well as magnitude of faults are changed for producing the datasets for training the neural network. The scaled conjugate gradient descent backpropagation algorithm is used as a learning algorithm. The inputs to the neural network are the current dataset under normal as well as fault conditions. The target matrix is prepared by considering the time duration of fault in a considered current signal. The pattern recognition tool in MATLAB is used as a training platform for neural network. A trained model is generated after training. This model is used in transmission line model for testing the fault conditions with the different magnitude and duration than training fault conditions. In this way the monitoring of faults is done in online mode. The obtained results of fault testings are presented in the paper

RNN Models for Dynamic Matrix Inversion: A Control-Theoretical Perspective

Abstract: In this paper, the existing recurrent neural network (RNN) models for solving zero-finding (e.g., matrix inversion) with time-varying parameters are revisited from the perspective of control and unified into a control-theoretical framework. Then, limitations on the activated functions of existing RNN models are pointed out and remedied with the aid of control-theoretical techniques. In addition, gradient-based RNNs, as the classical method for zero-finding, have been remolded to solve dynamic problems in manners free of errors and matrix inversions. Finally, computer simulations are conducted and analyzed to illustrate the efficacy and superiority of the modified RNN models designed from the perspective of control. The main contribution of this paper lies in the removal of the convex restriction and the elimination of the matrix inversion in existing RNN models for the dynamic matrix inversion. This work provides a systematic approach on exploiting control techniques to design RNN models for robustly and accurately solving algebraic equations.

Comprehensive Review on Detection and Classification of Power Quality Disturbances in Utility Grid With Renewable Energy Penetration

Abstract: The global concern with power quality is increasing due to the penetration of renewable energy (RE) sources to cater the energy demands and meet de-carbonization targets. Power quality (PQ) disturbances are found to be more predominant with RE penetration due to the variable outputs and interfacing converters. There is a need to recognize and mitigate PQ disturbances to supply clean power to the consumer. This article presents a critical review of techniques used for detection and classification PQ disturbances in the utility grid with renewable energy penetration. The broad perspective of this review paper is to provide various concepts utilized for extraction of the features to detect and classify the PQ disturbances even in the noisy environment. More than 220 research publications have been critically reviewed, classified and listed for quick reference of the engineers, scientists and academicians working in the power quality area.

Spatial Task Scheduling for Cost Minimization in Distributed Green Cloud Data Centers

Abstract: The infrastructure resources in distributed green cloud data centers (DGCDCs) are shared by multiple heterogeneous applications to provide flexible services to global users in a high-performance and low-cost way. It is highly challenging to minimize the total cost of a DGCDC provider in a market, where bandwidth prices of Internet service providers (ISPs), electricity prices, and the availability of renewable green energy all vary with geographical locations. Unlike existing studies, this paper proposes a spatial task scheduling and resource optimization (STSRO) method to minimize the total cost of their provider by cost-effectively scheduling all arriving tasks of heterogeneous applications to meet tasks’ delay-bound constraints. STSRO well exploits spatial diversity in DGCDCs. In each time slot, the cost minimization problem for DGCDCs is formulated as a constrained optimization one and solved by the proposed simulated annealing-based bat algorithm (SBA). Trace-driven experiments demonstrate that STSRO achieves lower total cost and higher throughput than two typical scheduling methods. Note to Practitioners —This paper investigates the cost minimization problem for DGCDCs while meeting delay-bound constraints for all arriving tasks. Previous task scheduling methods do not jointly investigate the spatial diversity in bandwidth prices of ISPs, electricity prices, and the availability of renewable green energy. Therefore, they fail to cost-effectively schedule all arriving tasks of heterogeneous applications during their delay-bound constraints. In this paper, a new method that overcomes the shortcomings of the existing methods is proposed. It is obtained by using the proposed SBA that solves a constrained optimization problem. Simulation results demonstrate that compared with two typical scheduling methods, it increases the throughput and decreases the cost. It can be readily implemented and integrated into real-world industrial DGCDCs. The future work needs to investigate the indeterminacy of renewable energy and the uncertainty in arriving tasks with rough deep neural network approaches on STSRO.